hectorjg/credit_card_marketing

Project: Credit Card Marketing

The objective of this project is to find the best prediction model on a classification data. The target is the Offer_accepted column. Here are 2 models, a linnera Regression model and a KN - neighbors model. The dataset used for this project can be found on GitHub-Ironhack

Objectives

• Create a Logistic Regression model to predict the customers who will accept the offer.
• Create a KN-neighbors model to predict the customers who will accept the offer.
• Visualize the dataframes to understand the most important data.
• Select the model with the highest accuracy score.

Project Steps:

Data Cleaning

• Drop the customer number column.
• Change the data types of the columns wich I consider.
• Check for null values and drop them.
• Plot the numerical column and remove outliers.
• Transform the numerical data in order to make it like a standart deviation.
• Plot the categorical columns and bucket the ones wich have a few values.

Data Analysis

• Selecting the columns that looks to be the most correlated to the target.
• Up sampling the data in order to balance it

Processing Data

• Scaling the numerical data.
• Use OneHotEncoder for the categorical data, excluding the target.

Logistic Regresion Model

• Create the logistic regression model to predict the Offer accepted.
• Plot a confusion matrix to visualize the prediction.

KN-neighbors

-Find the best "k value" for the model. -Create a KN-neighbors model to predict the Offer accepted. -Plot a confusion matrix to visualize the prediction.

Evaluate Accuracy

• score and ROC AUC;

Conclussion:

-The best model for this case is the Logistic Regression model because it had a better accuracy.

---

link to tableu https://public.tableau.com/app/profile/h.ctor.mondragon.reyes/viz/credit_card_marketing/Dashboard1?publish=yes

---

The files required for the project are in the data_files file:

• original_dataframe.csv
• clean_data.csv
• up_sampling_data.csv
• scaled_numerical_data.csv
• encoded_categorical_data.csv
• concatinated_final_data.csv