Logistic optimization: Delivery drivers location optimization with Causal Inference

Introduction

Work on clients’ data to help it understand the primary causes of unfulfilled requests as well as come up with solutions that recommend drivers locations that increase the fraction of complete orders.

Business overview

We will try to answer some interesting question that cannot be answered by just analyzing observational data alone.

These questions can be similar to:

• Given drivers are recommended to move 1km every 30 mins in a selected direction, what happens to the number of unfulfilled requests?

• If we assume we know the location of the next 20% of orders within 5km accuracy, what happens to the number of unfulfilled requests?

• Had we changed the time requirements to drivers operating time in the past, what fractions of orders could have been completed?

• If I increased the number of drivers by 10% cumulative per month, what fraction of orders can be completed?

objective

Work on clients’ data to help client understand the primary causes of unfulfilled requests as well as come up with solutions that recommend drivers locations that increase the fraction of complete orders. Since drivers are paid based on the number of requests they accept, the solution will help client business grow both in terms of client satisfaction and increased business.

Data Cleaning

Drop columns with empty entries
Drop rows with NaN entries
Merge the Two tables

Feature Engineering

Generate month, day, week day, and an hour from the trip start time column.
Calculate the driver proximity to the order using trip origin and driver location when the driver got the order which is given in lat and lng in the second table.
Calculate trip distance and trip duration and then trip speed.
I also used the API from https://api.weatherbit.io/v2.0/history/daily? to get the weather at a given location and time-stamp. Public, school, regional and national holidays are calculated from the trip start time.

Visualization

plot driver distance vs acceptance rate
plot latitude vs longitude of dirver location

Causal Inference

Used causalnex libary to build structural model
generarated some graphs

Folder Structure

maps

generated maps screenshots from this project are found here

screenshots

Folder caointing information about structural model of the causalnex network and visualization

notebooks

Caintains separate notebooks for the following purposes:

EDA on separate datsets
EDA on merged two dasets
EDA and feature engineering on merged datasets.
Causal Inference and Causal Graphs

scripts

Contains methods and functions for data cleaning and data extraction tasks.

Some screenshots from the EDA

Locations of destination for orders Locations of origin for orders

tests

unit tests for the methods found in scripts directory

Instalation

clone https://github.com/niyotham/Causal-Inference-Logistic-optimization.git
cd Causal-Inference-Logistic-optimization
pip install -r requirements.txt

Future Works

Finalize the causal inference model
Do more feature engineering and visualizations
Improving and implement Logistic optimization
Collaborate with domain expert thoughts to create a more meaningful causal graph.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.