Skip to content

Files

Latest commit

fce5bf1 · Apr 4, 2021

History

History

Transaction-data-analysis-and-prediction

Problem Description

Analyse the transactional data and Identify such factors which impacts the daily transaction volume. Also, predict the daily transaction volume for next 5 days.

Data Description

Daily Transactional data for each month i.e. Jan-2017 to Oct-2017 per day transactional Data

  • Jan - 41668
  • Feb - 54522
  • Mar - 156429
  • Apr - 187696
  • May - 231851
  • Jun - 318340
  • Jul - 385274
  • Aug - 423830
  • Sep - 463834
  • Oct - 472852

Solution

  • Data Preparation
    • Synchronize the column name in each dataset
    • Checked for the duplicate entries for transactions
    • Checked for the 'NA' values
    • Repalce the NA values witha appropriate values (e.g. Replace NA Credit_card_fee values to 0)
  • Exploratory Data Analysis
    • Univariate analysis
    • Bi-variate analysis
  • Combined the Monthly data in to one
  • Performed the EDA on Merged dataset (i.e. Combined data for all the months)
  • converted the data into timeseries
  • Built a Time-series model
    • PLotted time-series
    • Checked for the white noise
    • Performed ADF and KPSS test
  • Predicted the next 5 Days Transaction