blog banner

Revv cars achieved an average forecast accuracy ranging between 65% to 85%

Introduction

Revv is an India’s fastest growing self-drive car company with different options to meet all your self-drive needs – Hourly Rental, OPEN (multi-brand annual car subscription), One-way Outstation Rental, SWITCH (multi-brand monthly car subscription) and Hyundai Subscription (operated by Revv). Enjoy unique features – Prices with or without fuel, Flexibility to choose different delivery & return locations and Unlimited KMs option.

Why Pluto7?

Pluto7 Consulting Inc. and its affiliates (referred to as “Pluto7”), are an Artificial Intelligence (“Ai”), Machine Learning (“ML”) and Data Analytics Services Company with domain expertise in retail, supply-chain, manufacturing, hi-tech, higher education and healthcare bringing accurate business solutions.  

Challenges

Revv cars have limited machine learning mechanisms with no integration with external demand signals in place for Demand Forecasting to understand future demand on demand and other relevant factors. They calculate the total number of monthly hours a particular vehicle is available for booking. Out of these total hours, certain hours need to be deducted for service & maintenance or other services. The prime objective is to minimize the idle time of vehicles and predict total booking hours at city level. 

Results

Preparation of a Univariate dataset which is consistent with the structured data provided. Got external datasets of holiday and temperature for the city of Mumbai for the desired time period. Our team helped in building data studio dashboards for forecast done and  prepared training datasets for rolling forecast on a rolling window of 7

A rolling forecast was generated for 7 days with an average accuracy ranging between  65% to 85% with the actual booking happened during the specific time period was presented in 7 csv files.

For more information
www.pluto7.com/success-stories

industry image

Industry High-tech

Organization Name: Revv

Challenges

  • Gaps in data due to covid and other business reasons
  • Preparing multivariate dataset by incorporating external data
  • Analyze the impact of external features on predicting booking hours
  • Building Statistical, Machine Learning, Google’s Vertex AI Forecasting
  • Comparing performance of various models built and finalizing one model which performs better on all the car_ids

Results

  • Build data studio dashboards for forecast
  • Prepared training datasets for rolling forecast
  • Generated rolling forecast for 7 days with an average accuracy of ranging between 65% to 85%

Products Used

  • Google Cloud Storage
  • Big Query
  • Vertex AI Notebooks
  • Vertex AI forecasting
  • Looker