Análisis de Datos Segmentación de clientes potenciales (Customer Segmentation)¶
Comprension del negocio¶
INTRODUCCION
La segmentación de clientes es la práctica de dividir una base de clientes en grupos de individuos que son similares en aspectos específicos relevantes para el marketing, como la edad, el género, los intereses y los hábitos de gasto.
Las empresas que emplean la segmentación de clientes parten del hecho de que cada cliente es diferente, en preferencias y necesidades, y de que los esfuerzos de marketing se verían más favorecidos si se dirigen mensajes a grupos de segmentos específicos.
ENTORNO DE LA EMPRESA
Una empresa de consumo tiene previsto entrar en nuevos mercados con sus productos actuales (P1, P2, P3, P4 y P5). Tras un intenso estudio de mercado, han deducido que el comportamiento del nuevo mercado es similar al de su mercado actual.
En el mercado actual, el equipo de ventas ha clasificado a todos los clientes en 4 segmentos (A, B, C, D ). A continuación, han llevado a cabo una difusión y comunicación segmentada para un segmento diferente de clientes. Esta estrategia les ha funcionado excepcionalmente bien. Planean utilizar la misma estrategia para los clientes potenciales en los nuevos mercados.
OBJETIVO
En este proyecto de aprendizaje automático, nuestro objetivo es crear un modelo para predecir adecuadamente la Categorizacion/Clasificacion de los nuevos clientes.
Estructura del dataset¶
*El conjunto de datos contiene información de clientes en 11 variables, donde 9 variables son caracteristicas demograficas, que permite a la empresa clasificar en segmentos de clientes para fines de marketing.*
ID: Identificación única
Gender: Género del cliente
Ever_Married: Estado civil del cliente
Age: Edad del cliente
Graduated: ¿Es el cliente un graduado?
Profession: Profesión del cliente
Work_Experience: Experiencia laboral en años
Spending_Score: Puntuación del gasto del cliente
Family_Size: Número de miembros de la familia del cliente.
Var_1: Categoría anónima del cliente
Segmentation: (target) Segmento del cliente
Bibliotecas¶
Instalando librerías¶
In [1]:
%pip install --pre pycaret
Requirement already satisfied: pycaret in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (3.1.0) Requirement already satisfied: ipython>=5.5.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (8.7.0) Requirement already satisfied: ipywidgets>=7.6.5 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (8.1.1) Requirement already satisfied: tqdm>=4.62.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (4.66.1) Requirement already satisfied: numpy<1.24,>=1.21 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.23.5) Requirement already satisfied: pandas<2.0.0,>=1.3.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.5.3) Requirement already satisfied: jinja2>=1.2 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (3.1.2) Requirement already satisfied: scipy~=1.10.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.10.1) Requirement already satisfied: joblib>=1.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.2.0) Requirement already satisfied: scikit-learn<1.3.0,>=1.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.2.0) Requirement already satisfied: pyod>=1.0.8 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.1.0) Requirement already satisfied: imbalanced-learn>=0.8.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.11.0) Requirement already satisfied: category-encoders>=2.4.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.6.2) Requirement already satisfied: lightgbm>=3.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (4.1.0) Requirement already satisfied: numba>=0.55.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.58.0) Requirement already satisfied: requests>=2.27.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.31.0) Requirement already satisfied: psutil>=5.9.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (5.9.4) Requirement already satisfied: markupsafe>=2.0.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.1.1) Requirement already satisfied: importlib-metadata>=4.12.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (6.8.0) Requirement already satisfied: nbformat>=4.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (5.9.2) Requirement already satisfied: cloudpickle in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.2.1) Requirement already satisfied: deprecation>=2.1.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.1.0) Requirement already satisfied: xxhash in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (3.4.1) Requirement already satisfied: matplotlib>=3.3.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (3.8.0) Requirement already satisfied: scikit-plot>=0.3.7 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.3.7) Requirement already satisfied: yellowbrick>=1.4 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.5) Requirement already satisfied: plotly>=5.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (5.17.0) Requirement already satisfied: kaleido>=0.2.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.2.1) Requirement already satisfied: schemdraw==0.15 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.15) Requirement already satisfied: plotly-resampler>=0.8.3.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.9.1) Requirement already satisfied: statsmodels>=0.12.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.14.0) Requirement already satisfied: sktime!=0.17.1,!=0.17.2,!=0.18.0,<0.22.0,>=0.16.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (0.21.1) Requirement already satisfied: tbats>=1.1.3 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (1.1.3) Requirement already satisfied: pmdarima!=1.8.1,<3.0.0,>=1.8.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pycaret) (2.0.3) Requirement already satisfied: patsy>=0.5.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from category-encoders>=2.4.0->pycaret) (0.5.3) Requirement already satisfied: packaging in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from deprecation>=2.1.0->pycaret) (22.0) Requirement already satisfied: threadpoolctl>=2.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from imbalanced-learn>=0.8.1->pycaret) (3.1.0) Requirement already satisfied: zipp>=0.5 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from importlib-metadata>=4.12.0->pycaret) (3.17.0) Requirement already satisfied: backcall in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.2.0) Requirement already satisfied: decorator in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (5.1.1) Requirement already satisfied: jedi>=0.16 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.18.2) Requirement already satisfied: matplotlib-inline in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.1.6) Requirement already satisfied: pickleshare in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.7.5) Requirement already satisfied: prompt-toolkit<3.1.0,>=3.0.11 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (3.0.36) Requirement already satisfied: pygments>=2.4.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (2.13.0) Requirement already satisfied: stack-data in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.6.2) Requirement already satisfied: traitlets>=5 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (5.7.0) Requirement already satisfied: colorama in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipython>=5.5.0->pycaret) (0.4.6) Requirement already satisfied: comm>=0.1.3 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipywidgets>=7.6.5->pycaret) (0.1.4) Requirement already satisfied: widgetsnbextension~=4.0.9 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipywidgets>=7.6.5->pycaret) (4.0.9) Requirement already satisfied: jupyterlab-widgets~=3.0.9 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from ipywidgets>=7.6.5->pycaret) (3.0.9) Requirement already satisfied: contourpy>=1.0.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (1.1.0) Requirement already satisfied: cycler>=0.10 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (0.11.0) Requirement already satisfied: fonttools>=4.22.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (4.42.1) Requirement already satisfied: kiwisolver>=1.0.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (1.4.5) Requirement already satisfied: pillow>=6.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (10.0.0) Requirement already satisfied: pyparsing>=2.3.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (3.0.9) Requirement already satisfied: python-dateutil>=2.7 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from matplotlib>=3.3.0->pycaret) (2.8.2) Requirement already satisfied: fastjsonschema in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from nbformat>=4.2.0->pycaret) (2.18.0) Requirement already satisfied: jsonschema>=2.6 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from nbformat>=4.2.0->pycaret) (4.19.1) Requirement already satisfied: jupyter-core in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from nbformat>=4.2.0->pycaret) (5.1.0) Requirement already satisfied: llvmlite<0.42,>=0.41.0dev0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from numba>=0.55.0->pycaret) (0.41.0) Requirement already satisfied: pytz>=2020.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pandas<2.0.0,>=1.3.0->pycaret) (2022.6) Requirement already satisfied: tenacity>=6.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly>=5.0.0->pycaret) (8.2.3) Requirement already satisfied: dash<3.0.0,>=2.11.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly-resampler>=0.8.3.1->pycaret) (2.14.0) Requirement already satisfied: orjson<4.0.0,>=3.8.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly-resampler>=0.8.3.1->pycaret) (3.9.9) Requirement already satisfied: trace-updater>=0.0.8 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly-resampler>=0.8.3.1->pycaret) (0.0.9.1) Requirement already satisfied: tsdownsample==0.1.2 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly-resampler>=0.8.3.1->pycaret) (0.1.2) Requirement already satisfied: Cython!=0.29.18,!=0.29.31,>=0.29 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pmdarima!=1.8.1,<3.0.0,>=1.8.0->pycaret) (3.0.3) Requirement already satisfied: urllib3 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pmdarima!=1.8.1,<3.0.0,>=1.8.0->pycaret) (1.26.13) Requirement already satisfied: setuptools!=50.0.0,>=38.6.0 in c:\program files\windowsapps\pythonsoftwarefoundation.python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\site-packages (from pmdarima!=1.8.1,<3.0.0,>=1.8.0->pycaret) (65.5.0) Requirement already satisfied: six in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from pyod>=1.0.8->pycaret) (1.16.0) Requirement already satisfied: charset-normalizer<4,>=2 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from requests>=2.27.1->pycaret) (2.1.1) Requirement already satisfied: idna<4,>=2.5 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from requests>=2.27.1->pycaret) (3.4) Requirement already satisfied: certifi>=2017.4.17 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from requests>=2.27.1->pycaret) (2022.9.24) Requirement already satisfied: deprecated>=1.2.13 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from sktime!=0.17.1,!=0.17.2,!=0.18.0,<0.22.0,>=0.16.1->pycaret) (1.2.14) Requirement already satisfied: scikit-base<0.6.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from sktime!=0.17.1,!=0.17.2,!=0.18.0,<0.22.0,>=0.16.1->pycaret) (0.5.2) Requirement already satisfied: Flask<2.3.0,>=1.0.4 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (2.2.2) Requirement already satisfied: Werkzeug<2.3.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (2.2.2) Requirement already satisfied: dash-html-components==2.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (2.0.0) Requirement already satisfied: dash-core-components==2.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (2.0.0) Requirement already satisfied: dash-table==5.0.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (5.0.0) Requirement already satisfied: typing-extensions>=4.1.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (4.4.0) Requirement already satisfied: retrying in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (1.3.4) Requirement already satisfied: ansi2html in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (1.9.0rc1) Requirement already satisfied: nest-asyncio in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (1.5.6) Requirement already satisfied: wrapt<2,>=1.10 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from deprecated>=1.2.13->sktime!=0.17.1,!=0.17.2,!=0.18.0,<0.22.0,>=0.16.1->pycaret) (1.16.0rc1) Requirement already satisfied: parso<0.9.0,>=0.8.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jedi>=0.16->ipython>=5.5.0->pycaret) (0.8.3) Requirement already satisfied: attrs>=22.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jsonschema>=2.6->nbformat>=4.2.0->pycaret) (23.1.0) Requirement already satisfied: jsonschema-specifications>=2023.03.6 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jsonschema>=2.6->nbformat>=4.2.0->pycaret) (2023.7.1) Requirement already satisfied: referencing>=0.28.4 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jsonschema>=2.6->nbformat>=4.2.0->pycaret) (0.30.2) Requirement already satisfied: rpds-py>=0.7.1 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jsonschema>=2.6->nbformat>=4.2.0->pycaret) (0.10.3) Requirement already satisfied: wcwidth in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from prompt-toolkit<3.1.0,>=3.0.11->ipython>=5.5.0->pycaret) (0.2.5) Requirement already satisfied: platformdirs>=2.5 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jupyter-core->nbformat>=4.2.0->pycaret) (2.6.0) Requirement already satisfied: pywin32>=1.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from jupyter-core->nbformat>=4.2.0->pycaret) (305) Requirement already satisfied: executing>=1.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from stack-data->ipython>=5.5.0->pycaret) (1.2.0) Requirement already satisfied: asttokens>=2.1.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from stack-data->ipython>=5.5.0->pycaret) (2.2.1) Requirement already satisfied: pure-eval in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from stack-data->ipython>=5.5.0->pycaret) (0.2.2) Requirement already satisfied: itsdangerous>=2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from Flask<2.3.0,>=1.0.4->dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (2.1.2) Requirement already satisfied: click>=8.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from Flask<2.3.0,>=1.0.4->dash<3.0.0,>=2.11.0->plotly-resampler>=0.8.3.1->pycaret) (8.1.3) Note: you may need to restart the kernel to use updated packages.
In [2]:
%pip install plotly --upgrade
Requirement already satisfied: plotly in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (5.17.0) Requirement already satisfied: tenacity>=6.2.0 in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly) (8.2.3) Requirement already satisfied: packaging in c:\users\clabc\appdata\local\packages\pythonsoftwarefoundation.python.3.10_qbz5n2kfra8p0\localcache\local-packages\python310\site-packages (from plotly) (22.0) Note: you may need to restart the kernel to use updated packages.
Cargando librerias¶
Para el análisis de datos y el conjunto de datos¶
In [3]:
# Para instalar procesamiento de datos
import numpy as np
import pandas as pd
Para preprocesamiento y el modelo¶
In [4]:
#Importar la liberias de Pycaret
from pycaret.classification import *
Para gráficos¶
In [5]:
# Para instalar manejadores de graficos
import plotly.express as px #Liberia para graficos
Carga de Datos¶
Establecer la ruta de ubicacion del archivo con los datos
In [6]:
# Creación del dataframe (en todas las celdas de código que se requieran)
archivo_url = "https://raw.githubusercontent.com/Emilca/Pstg_UNI_Ciencia_Datos/main/Datasets/Customer_segmentation/customer_segmentation_train_old.csv"
*Creación del DataFrame para un conjunto de datos determinado, en este caso usaré el conjunto de datos "Segmentation"*
In [7]:
data = pd.read_csv(archivo_url) #los datos están separados por comas
print(data.head(5)) #vista del dataframe
ID Gender Ever_Married Age Graduated Profession Work_Experience \ 0 462809 Male No 22 No Healthcare 1.0 1 462643 Female Yes 38 Yes Engineer NaN 2 466315 Female Yes 67 Yes Engineer 1.0 3 461735 Male Yes 67 Yes Lawyer 0.0 4 462669 Female Yes 40 Yes Entertainment NaN Spending_Score Family_Size Var_1 Segmentation 0 Low 4.0 Cat_4 D 1 Average 3.0 Cat_4 A 2 Low 1.0 Cat_6 B 3 High 2.0 Cat_6 B 4 High 6.0 Cat_6 A
In [8]:
print(" ")
print(f"El dataframe tiene {list(data.shape)[0]} filas y {list(data.shape)[1]} columnas")
El dataframe tiene 8068 filas y 11 columnas
Preparando datos del dataframe¶
Info del dataset¶
In [9]:
print(data.info()) #Muestra los nombres de las columnas
<class 'pandas.core.frame.DataFrame'> RangeIndex: 8068 entries, 0 to 8067 Data columns (total 11 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 ID 8068 non-null int64 1 Gender 8068 non-null object 2 Ever_Married 7928 non-null object 3 Age 8068 non-null int64 4 Graduated 7990 non-null object 5 Profession 7944 non-null object 6 Work_Experience 7239 non-null float64 7 Spending_Score 8068 non-null object 8 Family_Size 7733 non-null float64 9 Var_1 7992 non-null object 10 Segmentation 8068 non-null object dtypes: float64(2), int64(2), object(7) memory usage: 693.5+ KB None
Verifica si hay valores NaN
In [10]:
print(data.isnull().sum())
ID 0 Gender 0 Ever_Married 140 Age 0 Graduated 78 Profession 124 Work_Experience 829 Spending_Score 0 Family_Size 335 Var_1 76 Segmentation 0 dtype: int64
---> Existen valores nulos y puede existir valores duplicados
Limpiando el dataset¶
Limpiando y rellenando con valores el dataset
In [11]:
#Limpiando el dataset con los valores más cercanos
#
#alguna vez casado
#data.fillna(value = {'Ever_Married':'N/A'})
data['Ever_Married'].fillna(method='ffill', inplace=True)
#alguna vez graduado
data['Graduated'].fillna(method='ffill', inplace=True)
#Profesion
data['Profession'].fillna(method='ffill', inplace=True)
#Experienccia laboral
data['Work_Experience'].fillna(method='ffill', inplace=True)
#Cantidad de familia
data['Family_Size'].fillna(method='ffill', inplace=True)
#Categoria
data['Var_1'].fillna(method='ffill', inplace=True)
Verificando si no hay valores nulos
In [12]:
print(data.isnull().sum())
ID 0 Gender 0 Ever_Married 0 Age 0 Graduated 0 Profession 0 Work_Experience 0 Spending_Score 0 Family_Size 0 Var_1 0 Segmentation 0 dtype: int64
Eliminando duplicados¶
In [13]:
data.drop_duplicates(inplace=True)
Eliminando Variable Id¶
Eliminar columna ID porque es innecesaria al modelo
In [14]:
data.drop(columns=["ID"], inplace=True) #Eliminado por nombre
#df = df.drop(axis=1, labels=['ID']) #Eliminado por numero de etiqueta
Crear Dataframe con datos limpios¶
Respaldamos el dataframe para preprocedar columnas manualmente y luego ingrearlo en pycaret
In [15]:
df = data.copy()
print(df)
Gender Ever_Married Age Graduated Profession Work_Experience \
0 Male No 22 No Healthcare 1.0
1 Female Yes 38 Yes Engineer 1.0
2 Female Yes 67 Yes Engineer 1.0
3 Male Yes 67 Yes Lawyer 0.0
4 Female Yes 40 Yes Entertainment 0.0
... ... ... ... ... ... ...
8063 Male No 22 No Artist 0.0
8064 Male No 35 No Executive 3.0
8065 Female No 33 Yes Healthcare 1.0
8066 Female No 27 Yes Healthcare 1.0
8067 Male Yes 37 Yes Executive 0.0
Spending_Score Family_Size Var_1 Segmentation
0 Low 4.0 Cat_4 D
1 Average 3.0 Cat_4 A
2 Low 1.0 Cat_6 B
3 High 2.0 Cat_6 B
4 High 6.0 Cat_6 A
... ... ... ... ...
8063 Low 7.0 Cat_1 D
8064 Low 4.0 Cat_4 D
8065 Low 1.0 Cat_6 D
8066 Low 4.0 Cat_6 B
8067 Average 3.0 Cat_4 B
[8068 rows x 10 columns]
Respaldamos Dataframe datos limpios¶
In [16]:
df.to_csv('data_segmentacion.csv', index=False)
Visualizacion y Analisis de variables¶
Visualización¶
Análisis bivariado Gráfico de correlación con las variables numéricas
In [17]:
# Calculate the correlation matrix
corr = round(df.corr(),4)
In [18]:
# Plot the heatmap
px.imshow(corr,
title = "Matriz de correlacion",
text_auto=True,
labels={"color":"Coeficiente"},
template="gridon")
Análisis bivariado Gráfico de correlación con las variables categóricas
In [19]:
categorical_features = ['Gender', 'Ever_Married', 'Graduated', 'Profession', 'Spending_Score', 'Var_1', 'Segmentation']
numerical_features = ['Age', 'Work_Experience', 'Family_Size']
In [20]:
corr_2 = round(df[categorical_features + numerical_features].corr(method='spearman'), 4)
In [21]:
# Plot the heatmap
px.imshow(corr_2,
title = "Matriz de correlacion",
text_auto=True,
labels={"color":"Coeficiente"},
template="gridon")
Distribucion de variables¶
In [22]:
for i in data.columns:
fig = px.histogram(data,
x = i,
template="gridon",
nbins=40)
fig.update_layout(bargap=0.2)
fig.show()
Ingeniería de Variables¶
En el conjunto de datos hay variables categóricas por lo que, antes de entrenar el modelo, es necesario aplicar ténicas de one-hot-encoding...
Transformación Gender-Ever_Married-Graduated¶
Las siguientes variables pueden ser codificadas en la etiqueta: Gender Ever_Married Graduated Spending_Score.
In [23]:
# Utilizamos Label encoding para las variables categoricas de dos valores (Si/No)
from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
columns = ['Gender','Ever_Married','Graduated']
for col in columns:
df[col] = le.fit_transform(df[col])
print(df)
Gender Ever_Married Age Graduated Profession Work_Experience \
0 1 0 22 0 Healthcare 1.0
1 0 1 38 1 Engineer 1.0
2 0 1 67 1 Engineer 1.0
3 1 1 67 1 Lawyer 0.0
4 0 1 40 1 Entertainment 0.0
... ... ... ... ... ... ...
8063 1 0 22 0 Artist 0.0
8064 1 0 35 0 Executive 3.0
8065 0 0 33 1 Healthcare 1.0
8066 0 0 27 1 Healthcare 1.0
8067 1 1 37 1 Executive 0.0
Spending_Score Family_Size Var_1 Segmentation
0 Low 4.0 Cat_4 D
1 Average 3.0 Cat_4 A
2 Low 1.0 Cat_6 B
3 High 2.0 Cat_6 B
4 High 6.0 Cat_6 A
... ... ... ... ...
8063 Low 7.0 Cat_1 D
8064 Low 4.0 Cat_4 D
8065 Low 1.0 Cat_6 D
8066 Low 4.0 Cat_6 B
8067 Average 3.0 Cat_4 B
[8068 rows x 10 columns]
Transformación Profession¶
Variables Profesion a aplicar One Hot Encoding
-- Revisar la distribucion de los datos para ver si los datos estan desbalanceados o no.
In [24]:
print(df['Profession'].value_counts())
Artist 2555 Healthcare 1353 Entertainment 963 Engineer 704 Doctor 703 Lawyer 636 Executive 605 Marketing 299 Homemaker 250 Name: Profession, dtype: int64
La variable "Profesión" tiene 9 categorías. Pero sólo las 3 iniciales son las de mayor frecuencia. Por lo tanto, el resto se agrupa en una nueva categoría denominada "Other".
In [25]:
df['Profession'] = df['Profession'].replace(['Lawyer','Executive','Marketing','Homemaker'],'Other')
In [26]:
#df = pd.get_dummies(df, columns = ['Profession'])
df['Profession'] = df['Profession'].map({'Artist': 0, 'Healthcare':1, 'Entertainment':2, 'Engineer': 3, 'Doctor':4, 'Other':5})
Transformación Var_1¶
Variables Var_1 a aplicar One Hot Encoding.
Revisar la distribucion de los datos para ver si los datos estan desbalanceados o no.
In [27]:
print(df['Var_1'].value_counts())
Cat_6 5287 Cat_4 1097 Cat_3 827 Cat_2 430 Cat_7 206 Cat_1 135 Cat_5 86 Name: Var_1, dtype: int64
La variable "Var_1" tiene 7 categorías. Pero sólo las 3 iniciales son las de mayor frecuencia. Por lo tanto, el resto se agrupa en una nueva categoría denominada "Other".
In [28]:
df['Var_1'] = df['Var_1'].replace(['Cat_5','Cat_1','Cat_7','Cat_2'],'Other')
print(df['Var_1'].value_counts())
Cat_6 5287 Cat_4 1097 Other 857 Cat_3 827 Name: Var_1, dtype: int64
Existen dos variables categóricas adicionales: Profesión y Var_1. Como son variables nominales, hay que apicarles la tecnica One Hot Encoding (Variables Dummies)
In [29]:
#df = pd.get_dummies(df, columns = ['Var_1'])
df['Var_1'] = df['Var_1'].map({'Cat_1': 0, 'Cat_2':1, 'Cat_3':2, 'Cat_4': 3, 'Cat_5':4, 'Cat_6':5, 'Cat_7':6})
Transformación 4-Spending_Score¶
Variables Spending_Score a aplicar transformación.
La variable 'Spending_Score' tiene tres valores : Low, Average and High. Como son variables ordinales, las codificare por separado.
In [30]:
df['Spending_Score'] = df['Spending_Score'].map({'Low': 0, 'Average':1, 'High':2})
Transformación 5-Segmentation¶
Variable Segmentation a aplicar transformación.
La variable 'Segmentation' tiene los valores : A, B, C, D. Como son variables ordinales, las codificare por separado.
In [31]:
df['Segmentation'] = df['Segmentation'].map({'A':0,'B':1,'C':2,'D':3})
In [32]:
print(df)
Gender Ever_Married Age Graduated Profession Work_Experience \
0 1 0 22 0 1 1.0
1 0 1 38 1 3 1.0
2 0 1 67 1 3 1.0
3 1 1 67 1 5 0.0
4 0 1 40 1 2 0.0
... ... ... ... ... ... ...
8063 1 0 22 0 0 0.0
8064 1 0 35 0 5 3.0
8065 0 0 33 1 1 1.0
8066 0 0 27 1 1 1.0
8067 1 1 37 1 5 0.0
Spending_Score Family_Size Var_1 Segmentation
0 0 4.0 3.0 3
1 1 3.0 3.0 0
2 0 1.0 5.0 1
3 2 2.0 5.0 1
4 2 6.0 5.0 0
... ... ... ... ...
8063 0 7.0 NaN 3
8064 0 4.0 3.0 3
8065 0 1.0 5.0 3
8066 0 4.0 5.0 1
8067 1 3.0 3.0 1
[8068 rows x 10 columns]
Correlación de las variables numéricas¶
In [33]:
#Creamos objeto de correlacion por Pearson
corr_matrix = round(df.select_dtypes(include=['float64', 'int']).corr(method='pearson'), 2)
In [34]:
px.imshow(corr_matrix,
title = "Matriz de correlacion",
text_auto=True,
labels={"color":"Coeficiente"},
template="gridon")
Análisis de EDA
¶
1. Entre las variables no existe mucha correlacion.¶
2. Existen valores NaN en columnas : Ever_Married,Graduated,Profession,Var_1,Work_Experience,Family_Size.¶
3. Existen 7 Variables string y 4 variables numericas.¶
4. La variable objetivo es multi categorica¶
5. Existen Profeson, Var_1 y Spending_Score multi categorica.¶
6. Existen 3 variables Si/No (Gender, Ever_Married, Graduated) y 4 variables multi categoricas.¶
7. Las variables que tiene mas correlacion con la variable objetivo que es segmento son Ever_Married, Age, Graduated, Spending_Score, Family_Size.¶
Observaciones:
Ever_Married - Los clientes no casados suelen estar en el segmento D mientras que los casados están en el segmento A, B o C.
Graduated - Los clientes graduados suelen estar en el segmento A, B o C mientras que los no graduados están en el segmento D.
Profession: Las profesiones de salud tienen peor califcacion y los artistas la mejor calificacion de segmento.
Spending_Score - Los clientes que gastan poco suelen estar en los
segmentos D, mientras que los que gastan mucho y de forma media están
en los segmentos A, B o C.
Work_Experience - Las mujeres entre mas experiencia laboral tienen se ubican en mejor segmento.
Ever_Married - Los clientes no casados suelen estar en el segmento D mientras que los casados están en el segmento A, B o C.
Graduated - Los clientes graduados suelen estar en el segmento A, B o C mientras que los no graduados están en el segmento D.
Profession: Las profesiones de salud tienen peor califcacion y los artistas la mejor calificacion de segmento.
Spending_Score - Los clientes que gastan poco suelen estar en los segmentos D, mientras que los que gastan mucho y de forma media están en los segmentos A, B o C.
Work_Experience - Las mujeres entre mas experiencia laboral tienen se ubican en mejor segmento.
Observaciones:
- Age: <30 están en el segmento D, 30-40 o >70 están en el segmento A mientras que 45-70 están en el segmento C
- Work_Experience: <2 están en el segmento C mientras que 6-11 están en el segmento A y D
- Family_Size: <1 están en el segmento A, 1-3 están en el segmento C y 4+ en el segmento D
Modelo de Clasificacion¶
Preparando el Dataframe¶
Recuperamos el dataset sin ingenieria de variables para generar un modelo con columnas originales.
In [35]:
#df_data = df.copy()
df_data = data.copy()
Agrupar la columna Var_1.
In [36]:
df_data['Var_1'] = df_data['Var_1'].replace(['Cat_5','Cat_1','Cat_7','Cat_2'],'Other')
Agrupar la columna Profession.
In [37]:
df_data['Profession'] = df_data['Profession'].replace(['Lawyer','Executive',
'Marketing','Homemaker'],'Other')
Columnas calculadas
Importante: Se valido en el modelo y no hubo mejoria en la probabilidad de prediccion.
In [38]:
df_data.head()
Out[38]:
| Gender | Ever_Married | Age | Graduated | Profession | Work_Experience | Spending_Score | Family_Size | Var_1 | Segmentation | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Male | No | 22 | No | Healthcare | 1.0 | Low | 4.0 | Cat_4 | D |
| 1 | Female | Yes | 38 | Yes | Engineer | 1.0 | Average | 3.0 | Cat_4 | A |
| 2 | Female | Yes | 67 | Yes | Engineer | 1.0 | Low | 1.0 | Cat_6 | B |
| 3 | Male | Yes | 67 | Yes | Other | 0.0 | High | 2.0 | Cat_6 | B |
| 4 | Female | Yes | 40 | Yes | Entertainment | 0.0 | High | 6.0 | Cat_6 | A |
Configuración de PyCaret¶
In [39]:
# Configurando PyCaret para trabajar (en todas las celdas de código que se requieran)
from pycaret.classification import *
Preparando train y test¶
Separamos los datos de entrenamiento y datos no vistos por el modelo que utilizaremos cuando probemos en produccion el modelo final
In [40]:
train = df_data.sample(frac=0.90, random_state=0)
test = df_data.drop(train.index)
Reseteamos los indices de las columnas
In [41]:
train.reset_index(inplace=True, drop=True)
test.reset_index(inplace=True, drop=True)
In [42]:
print("Datos para Modelar: " + str(train.shape))
print("Datos no vistos para Predicción: " + str(test.shape))
Datos para Modelar: (7261, 10) Datos no vistos para Predicción: (807, 10)
In [43]:
train.head()
Out[43]:
| Gender | Ever_Married | Age | Graduated | Profession | Work_Experience | Spending_Score | Family_Size | Var_1 | Segmentation | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Male | Yes | 72 | Yes | Other | 3.0 | High | 2.0 | Cat_6 | C |
| 1 | Male | No | 20 | No | Other | 1.0 | Low | 3.0 | Cat_4 | D |
| 2 | Male | Yes | 52 | Yes | Other | 7.0 | High | 4.0 | Cat_4 | C |
| 3 | Male | No | 47 | Yes | Entertainment | 2.0 | Low | 1.0 | Other | A |
| 4 | Male | Yes | 49 | Yes | Artist | 1.0 | Low | 4.0 | Cat_3 | C |
Guardar el archivo train y test con los datos preprocesados para utilizarlos en Power BI
In [44]:
train.to_csv('train_emp_pre.csv', index=False)
In [45]:
test.to_csv('test_emp_pre.csv', index=False)
Configurar setup del modelo¶
Ejecutamos setup de Pycaret, proponiendo como variable dependiente Segmentation. Dejaremos que Pycaret transforme las variables categoricas con transformacion Dummies o Onehot Encoding
In [46]:
cat_features = ['Gender', 'Ever_Married', 'Graduated', 'Profession', 'Spending_Score', 'Var_1']
In [47]:
modelo_setup = setup(data = train, target = "Segmentation", train_size = 0.7, session_id = 0,
#categorical_features = cat_features,
normalize = True,
imputation_type="iterative",
transformation = True
)
#,feature_selection = True #Elimina caracteristicas por debajo del umbral de correlacion.
#fix_imbalance = True , will automaticaaly fix the imbalanced dataset by oversampling using the SMOTE method.
#ignore_low_variance = True
#,multicollinearity_threshold = True #Evalua la exactitud de la prediccion.
#multicollinearity_threshold = 0.95 as inter-correlaciones más altas que el umbral definido se eliminan
#remove_multicollinearity = True,
#remove_outliers = True
| Description | Value | |
|---|---|---|
| 0 | Session id | 0 |
| 1 | Target | Segmentation |
| 2 | Target type | Multiclass |
| 3 | Target mapping | A: 0, B: 1, C: 2, D: 3 |
| 4 | Original data shape | (7261, 10) |
| 5 | Transformed data shape | (7261, 20) |
| 6 | Transformed train set shape | (5082, 20) |
| 7 | Transformed test set shape | (2179, 20) |
| 8 | Ordinal features | 3 |
| 9 | Numeric features | 3 |
| 10 | Categorical features | 6 |
| 11 | Preprocess | True |
| 12 | Imputation type | iterative |
| 13 | Iterative imputation iterations | 5 |
| 14 | Numeric iterative imputer | lightgbm |
| 15 | Categorical iterative imputer | lightgbm |
| 16 | Maximum one-hot encoding | 25 |
| 17 | Encoding method | None |
| 18 | Transformation | True |
| 19 | Transformation method | yeo-johnson |
| 20 | Normalize | True |
| 21 | Normalize method | zscore |
| 22 | Fold Generator | StratifiedKFold |
| 23 | Fold Number | 10 |
| 24 | CPU Jobs | -1 |
| 25 | Use GPU | False |
| 26 | Log Experiment | False |
| 27 | Experiment Name | clf-default-name |
| 28 | USI | 7bb0 |
In [48]:
modelo_setup.X_train_transformed
Out[48]:
| Gender | Ever_Married | Age | Graduated | Profession_Artist | Profession_Other | Profession_Engineer | Profession_Healthcare | Profession_Entertainment | Profession_Doctor | Work_Experience | Spending_Score_Low | Spending_Score_High | Spending_Score_Average | Family_Size | Var_1_Cat_6 | Var_1_Cat_3 | Var_1_Cat_4 | Var_1_Other | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | -1.103742 | -1.181541 | 0.144527 | 0.792998 | 1.488093 | -0.541415 | -0.307108 | -0.452273 | -0.365881 | -0.310929 | -0.073726 | 0.802245 | -0.420959 | -0.563564 | -1.534408 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 1 | 0.906009 | 0.846352 | 1.510256 | -1.261037 | -0.672001 | 1.847013 | -0.307108 | -0.452273 | -0.365881 | -0.310929 | -1.188259 | -1.246503 | 2.375529 | -0.563564 | -0.443331 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 2 | -1.103742 | 0.846352 | 0.083278 | -1.261037 | -0.672001 | -0.541415 | 3.256185 | -0.452273 | -0.365881 | -0.310929 | -0.073726 | 0.802245 | -0.420959 | -0.563564 | 0.301500 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 3 | -1.103742 | -1.181541 | -0.315483 | 0.792998 | -0.672001 | 1.847013 | -0.307108 | -0.452273 | -0.365881 | -0.310929 | -1.188259 | 0.802245 | -0.420959 | -0.563564 | -0.443331 | -1.379254 | 2.912275 | -0.396031 | -0.339783 |
| 4 | 0.906009 | -1.181541 | 0.263145 | 0.792998 | 1.488093 | -0.541415 | -0.307108 | -0.452273 | -0.365881 | -0.310929 | -0.073726 | 0.802245 | -0.420959 | -0.563564 | -0.443331 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 5077 | 0.906009 | 0.846352 | 0.020656 | 0.792998 | -0.672001 | -0.541415 | -0.307108 | 2.211054 | -0.365881 | -0.310929 | -0.073726 | -1.246503 | -0.420959 | 1.774421 | 0.863047 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 5078 | 0.906009 | -1.181541 | -0.781359 | 0.792998 | -0.672001 | -0.541415 | -0.307108 | -0.452273 | 2.733130 | -0.310929 | -1.188259 | 0.802245 | -0.420959 | -0.563564 | 0.863047 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 5079 | 0.906009 | -1.181541 | -0.617179 | -1.261037 | 1.488093 | -0.541415 | -0.307108 | -0.452273 | -0.365881 | -0.310929 | 0.737776 | 0.802245 | -0.420959 | -0.563564 | 2.001204 | -1.379254 | -0.343374 | 2.525056 | -0.339783 |
| 5080 | 0.906009 | -1.181541 | -0.538614 | -1.261037 | -0.672001 | -0.541415 | -0.307108 | -0.452273 | 2.733130 | -0.310929 | -0.073726 | 0.802245 | -0.420959 | -0.563564 | -0.443331 | 0.725030 | -0.343374 | -0.396031 | -0.339783 |
| 5081 | -1.103742 | 0.846352 | 0.486315 | 0.792998 | -0.672001 | -0.541415 | 3.256185 | -0.452273 | -0.365881 | -0.310929 | -0.073726 | -1.246503 | 2.375529 | -0.563564 | 1.311632 | -1.379254 | 2.912275 | -0.396031 | -0.339783 |
5082 rows × 19 columns
Comparar modelos¶
Procesamos para encontrar el mejor modelo
In [49]:
mejor_modelo = compare_models(n_select=3, fold = 10 ) #Comparamos modelos para saber cual crear Ejemnplo: , sort = 'Recall'
| Initiated | . . . . . . . . . . . . . . . . . . | 10:01:26 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Fitting 10 Folds |
| Estimator | . . . . . . . . . . . . . . . . . . | Logistic Regression |
| Model | Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | TT (Sec) | |
|---|---|---|---|---|---|---|---|---|---|
| gbc | Gradient Boosting Classifier | 0.5218 | 0.7844 | 0.5218 | 0.5117 | 0.5139 | 0.3590 | 0.3604 | 4.1610 |
| ada | Ada Boost Classifier | 0.5100 | 0.7537 | 0.5100 | 0.5012 | 0.5011 | 0.3434 | 0.3455 | 1.0300 |
| lightgbm | Light Gradient Boosting Machine | 0.5075 | 0.7747 | 0.5075 | 0.4993 | 0.5014 | 0.3402 | 0.3412 | 1.9150 |
| lr | Logistic Regression | 0.5053 | 0.7644 | 0.5053 | 0.4890 | 0.4896 | 0.3364 | 0.3399 | 2.9430 |
| ridge | Ridge Classifier | 0.5006 | 0.0000 | 0.5006 | 0.4810 | 0.4693 | 0.3285 | 0.3369 | 0.5040 |
| lda | Linear Discriminant Analysis | 0.4971 | 0.7613 | 0.4971 | 0.4921 | 0.4881 | 0.3269 | 0.3302 | 0.6840 |
| nb | Naive Bayes | 0.4850 | 0.7487 | 0.4850 | 0.4682 | 0.4670 | 0.3094 | 0.3141 | 0.5670 |
| rf | Random Forest Classifier | 0.4784 | 0.7451 | 0.4784 | 0.4760 | 0.4762 | 0.3022 | 0.3026 | 1.5960 |
| knn | K Neighbors Classifier | 0.4734 | 0.7192 | 0.4734 | 0.4848 | 0.4766 | 0.2975 | 0.2986 | 0.9540 |
| qda | Quadratic Discriminant Analysis | 0.4725 | 0.7408 | 0.4725 | 0.5134 | 0.4739 | 0.2976 | 0.3046 | 0.7930 |
| svm | SVM - Linear Kernel | 0.4723 | 0.0000 | 0.4723 | 0.4491 | 0.4442 | 0.2906 | 0.2979 | 0.6730 |
| et | Extra Trees Classifier | 0.4606 | 0.7159 | 0.4606 | 0.4588 | 0.4588 | 0.2787 | 0.2791 | 1.6280 |
| dt | Decision Tree Classifier | 0.4244 | 0.6220 | 0.4244 | 0.4294 | 0.4260 | 0.2316 | 0.2319 | 0.5290 |
| dummy | Dummy Classifier | 0.2820 | 0.5000 | 0.2820 | 0.0795 | 0.1240 | 0.0000 | 0.0000 | 0.5340 |
Crear Modelo¶
Creamos el modelo en base a la comparacion hecha por el metodo setup
In [50]:
modelo = create_model("lr")
#print(modelo)
| Initiated | . . . . . . . . . . . . . . . . . . | 10:04:49 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Fitting 10 Folds |
| Estimator | . . . . . . . . . . . . . . . . . . | Logistic Regression |
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4833 | 0.7403 | 0.4833 | 0.4680 | 0.4705 | 0.3061 | 0.3083 |
| 1 | 0.5187 | 0.7755 | 0.5187 | 0.4987 | 0.5019 | 0.3547 | 0.3583 |
| 2 | 0.5374 | 0.7730 | 0.5374 | 0.5243 | 0.5238 | 0.3801 | 0.3836 |
| 3 | 0.5059 | 0.7682 | 0.5059 | 0.4883 | 0.4894 | 0.3367 | 0.3401 |
| 4 | 0.5197 | 0.7738 | 0.5197 | 0.4967 | 0.5004 | 0.3560 | 0.3603 |
| 5 | 0.5059 | 0.7795 | 0.5059 | 0.4970 | 0.4875 | 0.3367 | 0.3419 |
| 6 | 0.5039 | 0.7730 | 0.5039 | 0.4869 | 0.4896 | 0.3346 | 0.3374 |
| 7 | 0.4665 | 0.7412 | 0.4665 | 0.4338 | 0.4436 | 0.2842 | 0.2880 |
| 8 | 0.4902 | 0.7564 | 0.4902 | 0.4859 | 0.4819 | 0.3175 | 0.3205 |
| 9 | 0.5217 | 0.7631 | 0.5217 | 0.5103 | 0.5078 | 0.3577 | 0.3611 |
| Mean | 0.5053 | 0.7644 | 0.5053 | 0.4890 | 0.4896 | 0.3364 | 0.3399 |
| Std | 0.0198 | 0.0134 | 0.0198 | 0.0233 | 0.0208 | 0.0266 | 0.0268 |
Optimizar el modelo¶
Optimizamos el modelo seleccionado considerando el tipo de optimizacion deseada.
In [51]:
modelo_optimizado = tune_model(modelo, fold=10, optimize = 'Precision')
#print(modelo_optimizado)
| Initiated | . . . . . . . . . . . . . . . . . . | 10:04:57 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Searching Hyperparameters |
| Estimator | . . . . . . . . . . . . . . . . . . | Logistic Regression |
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4754 | 0.7394 | 0.4754 | 0.4685 | 0.4711 | 0.2975 | 0.2978 |
| 1 | 0.5167 | 0.7753 | 0.5167 | 0.5106 | 0.5115 | 0.3536 | 0.3548 |
| 2 | 0.5315 | 0.7721 | 0.5315 | 0.5267 | 0.5266 | 0.3734 | 0.3748 |
| 3 | 0.5059 | 0.7684 | 0.5059 | 0.4966 | 0.4994 | 0.3385 | 0.3394 |
| 4 | 0.5177 | 0.7731 | 0.5177 | 0.5131 | 0.5116 | 0.3550 | 0.3573 |
| 5 | 0.5098 | 0.7798 | 0.5098 | 0.5056 | 0.5003 | 0.3436 | 0.3469 |
| 6 | 0.5059 | 0.7730 | 0.5059 | 0.4959 | 0.4987 | 0.3386 | 0.3398 |
| 7 | 0.4508 | 0.7406 | 0.4508 | 0.4373 | 0.4423 | 0.2652 | 0.2661 |
| 8 | 0.5000 | 0.7558 | 0.5000 | 0.5065 | 0.4999 | 0.3321 | 0.3339 |
| 9 | 0.5079 | 0.7628 | 0.5079 | 0.4988 | 0.5010 | 0.3405 | 0.3416 |
| Mean | 0.5022 | 0.7640 | 0.5022 | 0.4959 | 0.4962 | 0.3338 | 0.3352 |
| Std | 0.0219 | 0.0136 | 0.0219 | 0.0242 | 0.0223 | 0.0294 | 0.0297 |
Fitting 10 folds for each of 10 candidates, totalling 100 fits
Ensamblado de modelos¶
Utilizamos los métodos de ensamble de modelos para ayudar a mejorar el rendimiento de los modelos al mejorar su precisión.
-Bagging¶
- Multiples Modelos con distintos datos
In [52]:
bagged_modelo = ensemble_model(modelo_optimizado, method = 'Bagging',)
| Initiated | . . . . . . . . . . . . . . . . . . | 10:06:05 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Fitting 10 Folds |
| Estimator | . . . . . . . . . . . . . . . . . . | Bagging Classifier |
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4715 | 0.7396 | 0.4715 | 0.4637 | 0.4665 | 0.2920 | 0.2925 |
| 1 | 0.5029 | 0.7756 | 0.5029 | 0.4931 | 0.4950 | 0.3349 | 0.3367 |
| 2 | 0.5236 | 0.7703 | 0.5236 | 0.5199 | 0.5197 | 0.3630 | 0.3642 |
| 3 | 0.5020 | 0.7674 | 0.5020 | 0.4935 | 0.4966 | 0.3333 | 0.3338 |
| 4 | 0.5098 | 0.7717 | 0.5098 | 0.5072 | 0.5046 | 0.3448 | 0.3471 |
| 5 | 0.5059 | 0.7797 | 0.5059 | 0.5008 | 0.4971 | 0.3382 | 0.3410 |
| 6 | 0.5020 | 0.7742 | 0.5020 | 0.4886 | 0.4924 | 0.3330 | 0.3345 |
| 7 | 0.4685 | 0.7383 | 0.4685 | 0.4548 | 0.4598 | 0.2886 | 0.2896 |
| 8 | 0.4902 | 0.7545 | 0.4902 | 0.4942 | 0.4894 | 0.3187 | 0.3202 |
| 9 | 0.5157 | 0.7628 | 0.5157 | 0.5063 | 0.5080 | 0.3510 | 0.3524 |
| Mean | 0.4992 | 0.7634 | 0.4992 | 0.4922 | 0.4929 | 0.3298 | 0.3312 |
| Std | 0.0169 | 0.0139 | 0.0169 | 0.0187 | 0.0171 | 0.0227 | 0.0230 |
-Boosting¶
- Multiples Modelos y aprende del anterior. No utilizando porque no hay pesos en clases Target
In [53]:
boosted_modelo = ensemble_model(modelo_optimizado, method = 'Boosting')
#Estimator not supported for the Boosting method. Change the estimator or method to 'Bagging'.
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4558 | 0.7212 | 0.4558 | 0.4498 | 0.4523 | 0.2715 | 0.2717 |
| 1 | 0.5029 | 0.7612 | 0.5029 | 0.5016 | 0.5017 | 0.3357 | 0.3360 |
| 2 | 0.5118 | 0.7605 | 0.5118 | 0.5053 | 0.5072 | 0.3471 | 0.3478 |
| 3 | 0.4961 | 0.7578 | 0.4961 | 0.4898 | 0.4924 | 0.3257 | 0.3259 |
| 4 | 0.5118 | 0.7576 | 0.5118 | 0.5137 | 0.5101 | 0.3471 | 0.3485 |
| 5 | 0.4843 | 0.7648 | 0.4843 | 0.4765 | 0.4779 | 0.3091 | 0.3102 |
| 6 | 0.4902 | 0.7579 | 0.4902 | 0.4819 | 0.4850 | 0.3174 | 0.3178 |
| 7 | 0.4567 | 0.7356 | 0.4567 | 0.4492 | 0.4521 | 0.2733 | 0.2737 |
| 8 | 0.4843 | 0.7489 | 0.4843 | 0.4896 | 0.4853 | 0.3109 | 0.3118 |
| 9 | 0.4961 | 0.7453 | 0.4961 | 0.4911 | 0.4924 | 0.3252 | 0.3257 |
| Mean | 0.4890 | 0.7511 | 0.4890 | 0.4848 | 0.4857 | 0.3163 | 0.3169 |
| Std | 0.0188 | 0.0130 | 0.0188 | 0.0205 | 0.0193 | 0.0253 | 0.0254 |
-Blending¶
- Multiples Modelos para crear votacion
In [54]:
blend_soft_modelo = blend_models(estimator_list = [modelo_optimizado], method = 'soft')
| Initiated | . . . . . . . . . . . . . . . . . . | 10:06:29 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Fitting 10 Folds |
| Estimator | . . . . . . . . . . . . . . . . . . | Voting Classifier |
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4754 | 0.7394 | 0.4754 | 0.4685 | 0.4711 | 0.2975 | 0.2978 |
| 1 | 0.5167 | 0.7753 | 0.5167 | 0.5106 | 0.5115 | 0.3536 | 0.3548 |
| 2 | 0.5315 | 0.7721 | 0.5315 | 0.5267 | 0.5266 | 0.3734 | 0.3748 |
| 3 | 0.5059 | 0.7684 | 0.5059 | 0.4966 | 0.4994 | 0.3385 | 0.3394 |
| 4 | 0.5177 | 0.7731 | 0.5177 | 0.5131 | 0.5116 | 0.3550 | 0.3573 |
| 5 | 0.5098 | 0.7798 | 0.5098 | 0.5056 | 0.5003 | 0.3436 | 0.3469 |
| 6 | 0.5059 | 0.7730 | 0.5059 | 0.4959 | 0.4987 | 0.3386 | 0.3398 |
| 7 | 0.4508 | 0.7406 | 0.4508 | 0.4373 | 0.4423 | 0.2652 | 0.2661 |
| 8 | 0.5000 | 0.7558 | 0.5000 | 0.5065 | 0.4999 | 0.3321 | 0.3339 |
| 9 | 0.5079 | 0.7628 | 0.5079 | 0.4988 | 0.5010 | 0.3405 | 0.3416 |
| Mean | 0.5022 | 0.7640 | 0.5022 | 0.4959 | 0.4962 | 0.3338 | 0.3352 |
| Std | 0.0219 | 0.0136 | 0.0219 | 0.0242 | 0.0223 | 0.0294 | 0.0297 |
In [55]:
#blend_hard_modelo = blend_models(estimator_list = [modelo_optimizado], method = 'hard')
-Stacking¶
- Multiples Modelos como input del modelo final
In [56]:
stacking_soft_model = stack_models(estimator_list = [modelo_optimizado])
| Initiated | . . . . . . . . . . . . . . . . . . | 10:06:37 |
|---|---|---|
| Status | . . . . . . . . . . . . . . . . . . | Fitting 10 Folds |
| Estimator | . . . . . . . . . . . . . . . . . . | Stacking Classifier |
| Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|
| Fold | |||||||
| 0 | 0.4931 | 0.7509 | 0.4931 | 0.4824 | 0.4832 | 0.3194 | 0.3214 |
| 1 | 0.5344 | 0.7788 | 0.5344 | 0.5189 | 0.5200 | 0.3757 | 0.3790 |
| 2 | 0.5354 | 0.7771 | 0.5354 | 0.5257 | 0.5224 | 0.3776 | 0.3820 |
| 3 | 0.5217 | 0.7764 | 0.5217 | 0.5037 | 0.5060 | 0.3579 | 0.3611 |
| 4 | 0.5354 | 0.7788 | 0.5354 | 0.5140 | 0.5148 | 0.3767 | 0.3820 |
| 5 | 0.5098 | 0.7838 | 0.5098 | 0.4931 | 0.4920 | 0.3419 | 0.3465 |
| 6 | 0.5098 | 0.7807 | 0.5098 | 0.4956 | 0.4938 | 0.3422 | 0.3462 |
| 7 | 0.4862 | 0.7430 | 0.4862 | 0.4618 | 0.4675 | 0.3105 | 0.3138 |
| 8 | 0.4980 | 0.7583 | 0.4980 | 0.4980 | 0.4904 | 0.3275 | 0.3314 |
| 9 | 0.5098 | 0.7713 | 0.5098 | 0.4933 | 0.4924 | 0.3415 | 0.3455 |
| Mean | 0.5134 | 0.7699 | 0.5134 | 0.4987 | 0.4982 | 0.3471 | 0.3509 |
| Std | 0.0170 | 0.0133 | 0.0170 | 0.0176 | 0.0165 | 0.0230 | 0.0235 |
*** El mejor emsamblando es: stacking_soft_model***
- Mejora la metrica Accurancy y Precision del modelo
In [57]:
modelo_ensamblado = stacking_soft_model
Graficar el modelo¶
Curva ROC
In [58]:
plot_model(modelo_ensamblado, plot = "auc" )
Precision Recall Curve
In [59]:
plot_model(modelo_ensamblado, plot = "pr")
Tuned_custom_model
In [60]:
plot_model(modelo_ensamblado)
Matriz de Confusión
In [61]:
plot_model(modelo_ensamblado, plot = "confusion_matrix")
Intepretación de la Matriz de Confusión - Confusion Matrix
TN y TP - Acurrancy: Se puede observar que hay mayor efectividad de Verdaderos positivos y Falsos Positivos en la prediccion en las clases A(0), B(2), C(2), D(3), los que son Verdadero Positivo.
Falso Negativos (FN) - Recall (Bajo Recall - Sesgo):
- La clase real "A" (532 muestras) tiende dificultad para predecir (distinguir), hacia la clase D(118), C(99) y B(48) en ese orden de prioridad.
- La Clase real "B" tiende dificultad para predecir (distinguir), hacia la clase C, A y D en ese orden de prioridad.
- La Clase real "C" tiende dificultad para predecir (distinguir), hacia la clase A, D y C en ese orden de prioridad.
- La clase real "D" tiende dificultad para predecir (distinguir), hacia la clase A, B, y C en ese orden de prioridad.
Falso Negativos (FP) - Precision (Baja Precision - Varianza):
- La prediccion clase A (685 muestras) tiende a ser menos confiable (seguro) hacia la clase B(117), D(142) y C(99) en ese orden.
- La prediccion Clase B tiende a ser menos confiable (seguro) hacia la clase A, C y D en ese orden.
- La prediccion Clase C tiende a ser menos confiable (seguro) hacia la clase B, A y D en ese orden.
- La prediccion clase D tiende a ser menos confiable (seguro) hacia la clase la clase B, D y C en ese orden.
Curva de aprendizaje
In [62]:
plot_model(modelo_ensamblado, plot = "learning")
Importancia variables
In [63]:
plot_model(modelo_optimizado, plot = "feature")
Threshold
- Importante: Como la varaible objetivo es tiene una Clasificación multiclase, (A,B,C,D), este grafico no se permite
In [64]:
#plot_model(modelo, plot = "threshold")
Evaluar el modelo¶
Analizar el rendimiento del mejor modelo entrenado
In [65]:
evaluate_model(modelo_ensamblado, fold = 10)
interactive(children=(ToggleButtons(description='Plot Type:', icons=('',), options=(('Pipeline Plot', 'pipelin…
Finalizar Modelo¶
In [66]:
modelo_final = finalize_model(modelo_ensamblado)
Predecir modelo¶
In [67]:
predict_test = predict_model(modelo_ensamblado, data = test)
#predict_model(modelo_optimizado)
Resultado de predicción
In [68]:
results = pull()
results.head()
Out[68]:
| Model | Accuracy | AUC | Recall | Prec. | F1 | Kappa | MCC | |
|---|---|---|---|---|---|---|---|---|
| 0 | Stacking Classifier | 0.5118 | 0.7744 | 0.5118 | 0.5105 | 0.4999 | 0.3494 | 0.3543 |
In [69]:
predict_test.head()
Out[69]:
| Gender | Ever_Married | Age | Graduated | Profession | Work_Experience | Spending_Score | Family_Size | Var_1 | Segmentation | prediction_label | prediction_score | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Male | No | 22 | No | Healthcare | 1.0 | Low | 4.0 | Cat_4 | 3 | D | 0.9000 |
| 1 | Female | No | 18 | No | Healthcare | 3.0 | Low | 4.0 | Cat_6 | 3 | D | 0.8767 |
| 2 | Female | No | 58 | No | Other | 1.0 | Average | 3.0 | Cat_3 | 1 | B | 0.3451 |
| 3 | Male | Yes | 56 | No | Artist | 1.0 | Average | 3.0 | Cat_6 | 2 | C | 0.5060 |
| 4 | Male | Yes | 56 | Yes | Other | 1.0 | High | 2.0 | Cat_6 | 0 | B | 0.3404 |
Guardamos los datos de predicción para resultado de la predicción
In [70]:
predict_test.to_excel("Datos_Prediccion_entrenamiento.xlsx")
Guardar Modelo¶
In [71]:
save_model(modelo_final, 'Modelo_Segmentacion_Customer')
Transformation Pipeline and Model Successfully Saved
Out[71]:
(Pipeline(memory=Memory(location=None),
steps=[('label_encoding',
TransformerWrapperWithInverse(exclude=None, include=None,
transformer=LabelEncoder())),
('iterative_imputer',
TransformerWrapper(exclude=None, include=None,
transformer=IterativeImputer(add_indicator=False,
cat_estimator=LGBMClassifier(boosting_type='gbdt',
class_weight=None,
colsample_bytree=1....
verbose=0,
warm_start=False))],
final_estimator=LogisticRegression(C=1.0,
class_weight=None,
dual=False,
fit_intercept=True,
intercept_scaling=1,
l1_ratio=None,
max_iter=1000,
multi_class='auto',
n_jobs=None,
penalty='l2',
random_state=0,
solver='lbfgs',
tol=0.0001,
verbose=0,
warm_start=False),
n_jobs=-1, passthrough=True,
stack_method='auto', verbose=0))],
verbose=False),
'Modelo_Segmentacion_Customer.pkl')
Produccion¶
Accediendo al modelo¶
In [72]:
saved_modelo = load_model("Modelo_Segmentacion_Customer")
Transformation Pipeline and Model Successfully Loaded
Creando registro de cliente para evaluar¶
In [73]:
cliente_nuevo = pd.DataFrame({'Gender' : ['Male'],
'Ever_Married' : ['Yes'],
'Age' : [40],
'Graduated' : ['Yes'],
'Profession' : ['Artist'],
'Work_Experience' : [10],
'Spending_Score' : ['High'],
'Family_Size' : [9],
'Var_1' : ['Cat_5']})
cliente_nuevo
Out[73]:
| Gender | Ever_Married | Age | Graduated | Profession | Work_Experience | Spending_Score | Family_Size | Var_1 | |
|---|---|---|---|---|---|---|---|---|---|
| 0 | Male | Yes | 40 | Yes | Artist | 10 | High | 9 | Cat_5 |
Mostrando la predicción¶
Ingresando los datos del cliente al modelo
In [74]:
nueva_prediccion = predict_model( saved_modelo, data = cliente_nuevo) #, probability_threshold = 0.4)
nueva_prediccion.head()
Out[74]:
| Gender | Ever_Married | Age | Graduated | Profession | Work_Experience | Spending_Score | Family_Size | Var_1 | prediction_label | prediction_score | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Male | Yes | 40 | Yes | Artist | 10 | High | 9 | Cat_5 | C | 0.5682 |
Conclusiones¶
Podemos concluir que las caracteristicas con mayor importancia en el modelo son: Profession. Family_size, Graduated y Spending_score
El utilizar el metodo ffill para el relleno de datos contribuyo a una considerable mejoria en el modelo.
Se seleccionar modelo lr - Logistic Regression" por lo siguiente:
- Enfocan en disminuir el sesgo.
- Se utiliza el mejor Acurrancy considerando que la variable objetivo tiene una distribucion equilibrada.
- *Se ajusta los datos de train / test a 80/20 con mejores resultados.*
- *La velocidad de procesamiento del modelo es de la mas baja..*
El modelo base es capaz de predecir en un 49 % de las observaciones del conjunto de datos de prueba y mejora a 50.al optimizar el modelo..
Los resultados del modelo no son tan buenos en las categorias B y C.
No hubo mejoría en el modelo al utilizar las Variable: Age, Work Experience y Family_size, para generar una categoría "bin" (es necesario realizar más comprobaciones).
Para crear un modelo más robusto es necesario: (1) más datos sobre los clientes, (2) si hay más características relevantes
Los datos son insuficientes para mejorar el modelo, como nos sugiere la grafica de Curva de aprendizaje.