Understand Your Customers: Will They Buy?

Predictive Insights at Your Fingertips

Overview

At Predictive Shopper Insights, we bring the power of machine learning to your fingertips, helping you understand and anticipate online shopping behaviors with precision. Our platform leverages state-of-the-art algorithms to analyze user inputs and predict whether a customer will make a purchase. Whether you're optimizing marketing strategies, enhancing user experience, or improving conversion rates, our predictive models provide invaluable insights to drive your business forward. Enter your data and discover the likelihood of customer purchases with ease and accuracy.

DataSet

Dataset Overview

This dataset consists of 12,330 sessions, collected over a year to avoid biases. 84.5% of sessions did not result in purchases.

Features

The dataset includes 17 features such as Administrative, ProductRelated, BounceRates, PageValues, and SpecialDay. These features help in predicting user purchasing intentions.

Target Variable

The target variable is Revenue, indicating whether the session ended with a purchase. This helps in training the machine learning models.

Data Collection

The data was collected from different users to ensure variability and avoid bias towards specific campaigns, special days, or user profiles.

Modeling Approach

Data Preprocessing

Load the dataset and inspect for any anomalies. Handle missing values, encode categorical variables, and scale/normalize numerical features.

Feature Engineering

Select relevant features based on domain knowledge and EDA analysis.

Model Selection

Train various machine learning models (Logistic Regression, Random Forest Classifier, SVM, Gradient Boosting, MLP).

Model Evaluation

Evaluate model performance using metrics such as accuracy, precision, recall, F1 score, and AUC-ROC. Use cross-validation to ensure robustness.

Final Model Selection

Based on evaluating metrics, the Random Forest Classifier was selected as the final model due to its superior performance and better output.

Deployment

Deploy our final model using a web framework (e.g., Flask) for real-time predictions. Integrate the model with an e-commerce web app to provide actionable insights.

Results Based on Our Final Model

Accuracy

Precision

Recall

F1 score

Team

Our working team

E-Com Shopping Purchase Predictor