Icon

AABA_​Group1

Workflow Overview

Prepared by Business Digitalisation AABA Group 1

Choo Li Yun S10259220H | Emily Choo Hope S10269237A | Tay Gilbert S10265887F | Kelvin Puthenveettil Vinu S10259168A |

1 - Data Ingestion
2 - Data Cleaning
3 - Data Validation & Feature Engineering
4 - EDA, why do orders go unfulfilled?
5 - Model Inputs, Profiling & Exports
4 - EDA, why do orders go unfulfilled?
7 - Train/Test Split
8 - Primary Decision Tree Model
11 - Business Output & Risk Bands
11 - Model Comparison
9 - Benchmark Model, Random Forest
Reads GoGoX's order requests
Excel Reader
Runs Prediction on Test Site
Decision Tree Predictor
Checks for duplicateOrder IDs
Duplicate Row Filter
Removes outlier"active" order
Nominal Value Row Filter
Strips +0800from Created at
String Manipulation
Strips +0800from Pickup Time
String Manipulation
Converts datatype to date-time
String to Date&Time
Scorer
Classify outcome type
Rule Engine
Orders and GMVby outcome
GroupBy
Cancellations byPick Up Hour Chart
Line Plot
Cancellations byLead Time Chart
Bar Chart
Missing Value
ROC Curve
Trains random forest(100 trees)
Random Forest Learner
Creates bar chart:Orders by Outcome
Bar Chart
Runs predictionson test set
Random Forest Predictor
Allows to View Tree
Decision Tree View
ROC: Random Forest
ROC Curve
Join DT and RF probabilities
Joiner
Drops unscoredrows
Missing Value
Scores accuracy & kappa
Scorer
Splits predictions into5 risk tiers
Binner
Extract pickup year
Date&Time Part Extractor
Renames toRisk_Tier
Column Renamer
ROC: DT vs RF
ROC Curve
Calculate bookinglead time (hrs)
Date&Time Difference
Extract pickuphour and day
Date&Time Part Extractor
Createslead-time tiers
Rule Engine
Removes negativelead times
Rule-based Row Filter
Keeps request-timefeatures only
Column Filter
Cancellations byPick Up Hour
GroupBy
Excludes one 2030 pickupas a suspected date error
Rule-based Row Filter
Actual Cancel Rateby Tier
Crosstab
Creates target:Cancelled (Yes/No)
Rule Engine
Creates 1/0cancel flag
Rule Engine
70/30 stratified split(seed 42)
Table Partitioner
Cancellations byVehicle Chart
Bar Chart
Trains Decision Tree(MDL pruned)
Decision Tree Learner
Profile model dataset
Statistics
Cancellations by vehicle
GroupBy
Saves & exportsmodel inputs
Excel Writer
Cancellations by Lead Time
GroupBy

Nodes

Extensions

Links