Introduction

In this project data from a food delivery app is analysed with the purpose of generating insights and creating a prediction model. The data contains 18706 observations and 12 features as listed below.

Data Overview

The original data contains metrics related to order delivery and select weather metrics. The list below shows variables/features contained in the raw data.

• ACTUAL_DELIVERY_MINUTES - ESTIMATED_DELIVERY_MINUTES
• ITEM_COUNT
• USER_LAT
• USER_LONG
• VENUE_LAT
• VENUE_LONG,
• ESTIMATED_DELIVERY_MINUTES
• ACTUAL_DELIVERY_MINUTES,
• CLOUD_COVERAGE
• TEMPERATURE
• WIND_SPEED
• PRECIPITATION

Feature Engineering

In general, we use the available features to generate extra features that would be useful in providing insights either extra insights or could be used to enhance modelling algorithms.

Features such as USER_LAT, USER_LONG, VENUE_LAT, VENUE_LONG are used to compute the distance between the user and venue,

• USER_VENUE_DIST (Distance between user and venue)
• DayOfWeek (Day of the week)
• user_coordinates, venue_coordinates (estimated user location)
• venue_postal_code, user_postal_code (estimated venue location)
• Orders per hour (Orders issued/received in a given hour
• source_of_order_pcode (A classification feature (IN/OUT) used to indicate whether an order came from within the postal code or an area outside the postal code

back to top