Geospatial Modelling, AI and Machine Learning

Spatial Modelling: Deterministic, Stochastic & Hybrid

SM

Watch the Video ↗

Access the Presentation ↗

  • Spatial modelling: concept & need; deterministic models (rule-based, fixed output)
  • Stochastic models: probabilistic output; Monte Carlo simulation; uncertainty in spatial predictions
  • Hybrid models: combining deterministic & stochastic approaches

Spatial Interpolation: IDW — Formula & Limitations

  • Spatial interpolation: concept; choosing interpolation method based on data type & distribution
  • IDW (Inverse Distance Weighting): formula, power parameter, advantages & limitations

Spatial Interpolation: Kriging & Variogram

  • Kriging: geostatistical interpolation; variogram; ordinary, universal & co-kriging; advantages

Spatial Interpolation: Spline — Regularized & Tension

  • Spline: regularized vs tension spline; applications in smooth surface generation

Spatial Regression: Spatial Lag & Spatial Error Models

  • Linear regression in spatial context: OLS; spatial autocorrelation of residuals (Moran’s I)

Logistic Regression: Binary Spatial Outcomes

  • Logistic regression: binary spatial outcomes (flood/no flood, diseased/healthy)
  • Spatial regression: spatial lag model; spatial error model; accounting for spatial dependency

Agent-Based Modelling: Concept & Applications

  • Agent-based modelling: concept; agents, rules, environment; emergent behaviour; applications in urban growth, land use change

Machine Learning Fundamentals: Types & Workflow

  • Machine learning fundamentals: definition; ML vs traditional statistics; ML workflow

Supervised Learning: Classification vs Regression

  • Supervised learning: labelled training data; classification vs regression tasks

Unsupervised Learning: Pattern Discovery & Clustering

  • Unsupervised learning: unlabelled data; pattern discovery; clustering

Reinforcement Learning: Concept in Geography

  • Reinforcement learning: reward-based learning; agents & environment; less common in geography

Random Forest: Ensemble Method & Feature Importance

  • Random Forest (RF): ensemble of decision trees; bagging; feature importance; OOB error

Support Vector Machine (SVM): Hyperplane & Kernel Trick

  • Support Vector Machine (SVM): hyperplane, kernel trick, margin maximization; applications in image classification

Decision Trees: Splitting Criteria & Pruning

  • Decision Trees: splitting criteria (Gini impurity, entropy/information gain); overfitting; pruning

K-means Clustering: Centroid-Based & Elbow Method

  • K-means clustering: centroid-based; elbow method for K selection; limitations

KNN: Distance-Based Classification

  • K-Nearest Neighbours (KNN): distance-based classification; choice of K; computational cost

Artificial Neural Networks: Perceptron, Layers & Backpropagation

  • Artificial Neural Networks (ANN): perceptron, hidden layers, activation functions, backpropagation

Deep Learning: CNN, RNN & Transfer Learning

  • Deep learning: CNN (for image classification), RNN/LSTM (for time series); transfer learning

PCA & LDA: Dimensionality Reduction in ML

  • PCA in ML: dimensionality reduction; principal components; explained variance
  • LDA (Linear Discriminant Analysis): supervised dimensionality reduction; class separation

Feature Selection & Feature Hashing

  • Feature hashing: encoding categorical variables; hash functions
  • Feature selection: filter, wrapper & embedded methods; reducing overfitting

Model Evaluation: Accuracy, Precision, Recall, F1-score

  • Model evaluation metrics: accuracy, precision, recall, F1-score — definitions & formulas

AUC-ROC Curve & Cross-Validation

  • AUC-ROC: concept of ROC curve; area under curve; threshold selection; multi-class extension
  • Cross-validation: k-fold, stratified k-fold; train-test split; avoiding data leakage

ML Applications: LULC Classification

  • ML applications in LULC: training sample collection, classifier comparison, temporal change detection

ML Applications: Flood Susceptibility Mapping

  • ML applications in flood susceptibility: multi-criteria features (slope, TWI, rainfall, soil); ensemble methods

ML Applications: Landslide Susceptibility Mapping

  • ML applications in landslide susceptibility: LR, RF, SVM comparison; validation with AUC-ROC

ML Applications: Forest Fire & Species Distribution

  • ML applications in forest fire prediction: fire weather indices, fuel load mapping
  • ML applications in species distribution: MaxEnt, RF; habitat suitability modelling

ML Applications: Urban Growth Prediction

  • ML applications in urban growth prediction: CA-Markov, SLEUTH, ANN; future scenario modelling

Big Data in Geography: 4 Vs & Hadoop Ecosystem

  • Big data in geography: volume, velocity, variety, veracity — 4 Vs; geospatial big data characteristics
  • Hadoop ecosystem: HDFS (storage), MapReduce (processing), YARN; use cases in large raster processing

Apache Spark & Geospatial Big Data Analytics

  • Apache Spark: in-memory computation; faster than MapReduce; Spark Spatial for geospatial analytics

Spatial Indexing: R-tree, Quad-tree & Geohash

  • Spatial indexing: R-tree, quad-tree, geohash — efficient spatial querying

Spatial Clustering & Hotspot Analysis

  • Spatial clustering: DBSCAN, HDBSCAN; hotspot analysis (Getis-Ord Gi*)
  • Spatial data mining: association rules, sequential patterns, spatial co-location patterns

IoT in Geography: Devices, Sensors & Drones

  • IoT in geography: concept; types of geospatial IoT devices — GPS trackers, weather stations, soil sensors, drones

IoT Protocols: MQTT, CoAP & LoRaWAN

  • IoT protocols: MQTT (lightweight pub-sub messaging), CoAP (constrained devices), LoRaWAN (long-range low-power)

IoT–GIS Integration: Smart City & Precision Agriculture

  • IoT–GIS integration: real-time dashboards; smart city applications; precision agriculture; flood monitoring

Practical: Interpolation, Regression, RF/SVM/ANN in Python, IoT GIS

  • Practical: IDW/Kriging/Spline interpolation in QGIS/ArcGIS; linear & logistic regression in Python (scikit-learn); spatial regression in GeoDa/R; agent-based modelling in NetLogo/Python; RF, SVM, ANN implementation in Python; spatial clustering & hotspot analysis; IoT sensor data visualization in GIS

Supplementary Topics: Geospatial Modelling, AI & ML

No specific topics listed.