Initialising engine…
Generating PDF report…
CashLens Dashboard
Treasury Intelligence
FY 2024 · SME Edition · Hybrid AI Ensemble · Synthetic Demo Data
Net Cash Flow
Historical + 90D ensemble forecast · 95% PI
90D 180D 1Y
Liquidity Position
Cumulative cash · ₹ crores
Monthly Inflow vs Outflow
Avg by month · last 12M
Weekly Cash Pattern
Avg net cash by weekday
90-Day Forecast
Hybrid Ensemble · XGBoost 45% · LSTM 35% · Prophet 20%
Ensemble Active — Coverage at 95% PI: 93.8%. Toggle models below. Run cash_flow_forecasting_engine.py in Colab for live data.
All Models vs Actual
90-day test window · toggle visibility
Ensemble XGBoost LSTM Prophet
Prediction Intervals
Bootstrap 95% CI · 30D
Forecast Residuals
Error distribution histogram
30-Day Rolling Forecast
Ensemble · confidence bands included
DateDayForecast ₹ Lower 95%Upper 95%PI WidthSignal
Model Performance
Comparative evaluation · ML, statistical & deep learning
Accuracy Metrics
MAPE Comparison
Lower is better · test set
Ensemble Weights
SLSQP-optimised distribution
XGBoost Feature Importance
Top 15 · gain-based · SHAP-aligned
LSTM Training Curve
Huber loss · 30 epochs · AdamW + ReduceLROnPlateau
Engine Configuration
Upload data · tune models · export outputs
Data Input
Drop bank transaction CSV
Required: date · cash_inflow · cash_outflow
Optional: rbi_repo_rate · usd_inr · cpi_index
Demo Mode — Showing 2,557 days of synthetic SME data. Upload CSV to replace. Run cash_flow_forecasting_engine.py in Google Colab for full ML training.
Hyperparameters
Pipeline Architecture
📥
Ingest
CSV/API/ERP
2500+ rows
🔧
Features
40+ lag/roll
calendar+macro
🌳
XGBoost
Walk-fwd CV
500 trees
🧠
LSTM
2L·128H
Huber loss
📈
Prophet
Multiplicative
+macro regs
Ensemble
SLSQP wts
Bootstrap PI