Skip to content
Chemometrics (PCA, PLS)Process Analytical Technology (PAT)Machine LearningDigital dashboardsAI-powered statistical interpretation
Raw sugar concentrationpH control (lime / sulfuric acid neutralisation)Temperature profilesDO limitationsAgitationCell growth kineticsNIR/Raman spectra from PAT sensors
Simulates 300 fermentation batches with realistic process parametersProcess variables: Sugar concentration, lime/H2SO4 doses, temperature, pH, DO, agitation, cell density, CO2, durationNIR spectra: 60 wavelengths (900-1700nm) with sugar and lactate absorption bandsOutputs: Lactate yield (%), concentration (g/L), and batch quality flagCreates:
PCA: 5 components for spectral dimensionality reduction and outlier detection (T²-Q diagnostics)PLS Regression: 5 latent variables to predict lactate concentration from NIR spectraPreprocessing: StandardScaler for spectral data normalizationSaves:
Data exploration and visualizationPCA score plots colored by yield/concentrationT²-Q residual plots for batch quality assessmentPLS model performance evaluation (R², RMSE)Prediction vs observed plots
📈 Overview:Key Performance Indicators and Distribution plots (yield, concentration, quality), etc.
🔬 Chemometrics (PCA):Interactive PC selection (PC1-3)
🤖 ML Model (PLS): Predicted vs Observed plot💡 AI Insights: Google Gemini integration for intelligent data analysis
🌿 Fermentation Insight Hub — Chemometrics for Lactic Acid Production
A Data Science & Digitalisation Project by Chadaporn Khaolumloet (Tangmo)
1. Theory Summary
2. Project Overview
Fermentation Insight Hub is a full end-to-end data science solution designed to simulate lactic acid fermentation through:
3. Business Context
Corbion specialises in fermentation-based biochemicals, especially lactic acid and its derivatives.
Industrial fermentation processes are influenced by:
Variability in these factors directly affects lactate yield (%) and final concentration (g/L).
This portfolio demonstrates how a Data Science knowledge can:
👉 apply chemometrics
👉 analyse fermentation health
👉 model process relationships
👉 detect deviations early
👉 build digital tools for R&D + Operations
Project Overall
4. Project Deliverables
4.1 Project Workflow
Step
File Name
What This File Does
1. Synthetic Data Generation
src/generate_synthetic_data.py
Creates fermentation batches + NIR spectra.
2. Utilities for Chemometrics
src/chemometrics_utils.py
Helper functions for PCA, PLS, scaling, spectral preprocessing, etc.
3. Model Training (Chemometrics)
src/train_models.py
Trains PCA/PLS models and saves them as .joblib.
4. Dataset Storage
data/lactic_fermentation_synthetic.csv
Full synthetic fermentation dataset.
5. Saved Chemometrics Models
data/chemometrics_models.joblib
Contains PCA model, scaler, PLS model, wavelength list, etc.
6. EDA Notebook (PCA + Visualisation)
notebooks/01_eda_chemometrics.ipynb
Does exploratory data analysis + initial PCA.
7. Dashboard / Web App
app/app.py
Streamlit app for uploading data, PCA scoring, PLS predictions.
There are no rows in this table
4.2 Deliverables
1. Data Generation (generate_synthetic_data.py)
generate_synthetic_data.py
lactic_fermentation_synthetic.csv
2. Model Training (train_models.py + chemometrics_utils.py)
train_models.py
chemometrics_utils.py
chemometrics_models.joblib
3. Exploratory Analysis (01_eda_chemometrics.ipynb)
01_eda_chemometrics.ipynb
4. Interactive Dashboard (app.py)
app.py
5 Tabs with comprehensive analytics:
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.