• Tara Dehdari
• Nishok Narayanan
• Muris Saab

In this project, we aim to enhance customer satisfaction in food delivery services by predicting whether a food order is satisfactory or not. We've explored various aspects of data analysis and machine learning models to achieve this goal.

We started by importing the dataset and conducted exploratory data analysis to gain insights into the dataset, such as:

• Analyzing the distribution of cuisine types and days of the week.
• Identifying top restaurants based on order frequency.
• Investigating the relationships between numerical variables.

Some preprocessing steps:

• Renamed columns for better readability.
• Dropped unnecessary columns.
• Handled missing values by converting "Not given" ratings to NaN and then dropping them.
• Categorical variables were one-hot encoded
• Data was split into 30% test set and 70% training set.

We built and evaluated three machine learning models:

• Random Forest Classifier: Achieved an accuracy of around 81%.
• Logistic Regression: Achieved an accuracy of approximately 83%.
• Support Vector Machine (SVM): Also achieved an accuracy of around 83%. This model performed best, showing promising results in classifying satisfactory and unsatisfactory orders.

To enhance customer satisfaction in food delivery services, we recommend using the Support Vector Machine (SVM) model for binary classification. It offers a balance between precision and recall, making it a valuable tool for improving service quality and the overall customer experience. Since the dataset is imbalanced we also recommend grabbing data with lower ratings as well.

The dataset used in this project was obtained from Kaggle. This dataset provides information about food orders and delivery from various restaurants. It is a valuable resource for analyzing customer preferences, restaurant performance, and delivery times in the context of a food delivery application.