#!/bin/bash # Simple test data validation for hyperparameter optimization DATA_FILE="$1" if [ -z "$DATA_FILE" ]; then echo "Usage: $0 " exit 1 fi echo "Validating test data: $DATA_FILE" echo "============================================================" # Check file exists if [ ! -f "$DATA_FILE" ]; then echo "❌ File not found: $DATA_FILE" exit 1 fi echo "✓ File exists" # Check file size FILE_SIZE=$(stat -f%z "$DATA_FILE" 2>/dev/null || stat -c%s "$DATA_FILE" 2>/dev/null) FILE_SIZE_MB=$((FILE_SIZE / 1024 / 1024)) echo "✓ File size: ${FILE_SIZE_MB} MB" if [ $FILE_SIZE_MB -lt 1 ]; then echo "⚠️ Warning: File size is very small (< 1 MB)" fi # Check file extension if [[ ! "$DATA_FILE" =~ \.parquet$ ]]; then echo "⚠️ Warning: File does not have .parquet extension" fi # Check read permissions if [ ! -r "$DATA_FILE" ]; then echo "❌ File is not readable" exit 1 fi echo "✓ File is readable" # Basic file info echo "" echo "File Details:" ls -lh "$DATA_FILE" # Training suitability assessment if [ $FILE_SIZE_MB -gt 100 ]; then SUITABILITY="EXCELLENT" elif [ $FILE_SIZE_MB -gt 10 ]; then SUITABILITY="GOOD" else SUITABILITY="MODERATE" fi echo "" echo "Training Suitability: $SUITABILITY" echo "" echo "✅ Data validation PASSED - Ready for hyperparameter optimization"