Home

Data Overview

What you'll see here

Dataset shape, data types, duplicate and missing-value checks, statistical summary, and the class imbalance that drives every later decision.

Training set
20,000 × 41
40 predictors + Target
Test set
5,000 × 41
held out until the end
Failure rate (train)
5.55%
1,110 failures
Missing values
36
V1: 18, V2: 18 (train) · 11 in test
Target balance (training set)

Class imbalance (5.55% failures) drives every modelling decision that follows.

Data quality checks
CheckResult
Data typesAll 40 predictors float64; Target int64
Duplicate rows (train / test)0 / 0
Missing cells (train / test)36 / 11
Value range across predictors-20.37 to 23.63
Mean IQR-outlier share1.25% of values
Statistical summary

First 8 of 40 predictors. All are continuous

float64
, roughly centred, and on comparable but not identical scales — so standardisation is mandatory.

VariableMeanStdMinMedianMax
V1-0.273.44-11.88-0.7515.49
V20.443.15-12.320.4713.09
V32.483.39-10.712.2617.09
V4-0.083.43-15.08-0.1413.24
V5-0.052.1-8.6-0.18.13
V6-12.04-10.23-16.98
V7-0.881.76-7.95-0.928.01
V8-0.553.3-15.66-0.3911.68
  • Means fall between -3.61 and 2.48; standard deviations between 1.65 and 5.5.
  • No categorical variables, so no encoding is required — only scaling.
  • Predictor values span -20.37 to 23.63.
AI: Data quality & class balance

Click Generate to produce an evidence-based observation, insight, recommendation, and business benefit for this section.