Have a question?
Message sent Close
4.65 out of 5
4.65
76 reviews on Udemy

How to easily use ANN for prediction mapping using GIS data?

First Simplified Step-by-Step Artificial Neural Network Methodology in R for Prediction Mapping using GIS Data
Instructor:
Dr. Omar AlThuwaynee
278 students enrolled
English [Auto]
With Step by step description we will be together facing the common software and code misleadings.
1. Produce training and testing data using automated tools in QGIS (Optional). Or jump this and using your own training/testing data directly.
2. Run NeuralNet function with training data and testing data. (use my QGIS tools as an option OR use your preferable data production technique directly)
3. Plot NN function network and get all the outputs like; Error rate, statistics, Pairwise and Generalized weight plot
4- Prediction and Validation Mapping Accuracy using AUC value of ROC plot
4. Produce and export prediction map using Raster data

Artificial Neural Network (ANN) is one of the advanced Artificial Intelligence (AI) component, through many applications, vary from social, medical and applied engineering, ANN proves high reliability and validity enhanced by multiple setting options.

Using ANN with Spatial data, increases the confidence in the obtained results, especially when it compare to regression or classification based techniques. as called by many researchers and academician especially in prediction mapping applications.

Together, step by step with “school-bus” speed, will cover the following points comprehensively (data, code and other materials are provided) using NeuralNet Package in R and Landslides data and thematics maps.

  1. Produce training and testing data using automated tools in QGIS OR SKIP THIS STEP AND USE YOUR OWN TRAINING AND TESTING DATA

  2. Run Neural net function with training data and testing data

    1. Plot NN function network

    2. Pairwise NN model results of Explanatories and Response Data

    3. Generalized Weights plot of Explanatories and Response Data

  3. Variables importance using NNET Package function

    1. Run NNET function

    2. Plot NNET function network

    3. Variables importance using NNET

    4. Sensitivity analysis of Explanatories and Response Data

  4. Run Neural net function for prediction with validation data

    1. Prediction Validation results with AUC value and ROC plot

  5. Produce prediction map using Raster data

    1. Import and process thematic maps like, resampling, stacking, categorical to numeric conversion.

    2. Run the compute (prediction function)

    3. Export final prediction map as raster.tif

Introduction

1
Course outlines
2
Expected Outcomes

ANN basic background and used packages

1
Introduction to ANN and used functions
2
Introduction to NuralNet package
3
Introduction Summary

Create training and testing data in QGIS work environment

1
Adding my developed tools to QGIS processing library
2
Create Land Cover map (convert string observations to numeric) in QGIS
3
Run the tools Step 1
4
Run the tools Step 2
5
Run the tools Step 3

Manage training and testing data in Excel

1
Excel work step 1
2
Excel work step 2

Introduction to code settings and data processıng in R studio environment

1
Outlines of the code contents
2
Working directory settings and data input
3
Convert Slope Aspect Categorical data into Numeric
4
Convert Land-cover Categorical data into Numeric
5
Data Scaling
6
Testing Data processing

Run ANN NeuralNet (nn) package and get results plots

1
Run NeuralNet (nn) function
2
Plot NeuralNet (nn) and get error estimation
3
Adding NN function prediction output to training data frame
4
How to convert values from scaled to original dataframe
5
Pairwise plot of training dataframe and function output
6
Generalized weight (GW) plot of training dataframe and function output

(optional) Run NNET package and plot outputs

1
Run NNET function and get variables importance plot
2
Plot NNET function network
3
Run Sensitivity test using NNET function

Prediction map processing using NeuralNet (nn) function

1
Run compute function (prediction function) and get cross tabulation results
2
Update dataframe and run the previous step again
3
Get cross tabulation for updated dataframe prediction
4
Run compute function (prediction) with testing data and get cross tabulation
5
Run ROC for function success and prediction rate results

Final Prediction map production and visualization using NeuralNet

1
Import raster files into R studio
2
Rasters processing (extents, resampling and stacking)
3
Scale Rasters stack data
4
Run compute (prediction) function for Rasters stack data
5
Produce final prediction Raster map
6
Export prediction raster map to QGIS

Code Conclusion and Summary

1
Code Conclusion and Summary
You can view and review the lecture materials indefinitely, like an on-demand channel.
Definitely! If you have an internet connection, courses on Udemy are available on any device at any time. If you don't have an internet connection, some instructors also let their students download course lectures. That's up to the instructor though, so make sure you get on their good side!
4.7
4.7 out of 5
76 Ratings

Detailed Rating

Stars 5
36
Stars 4
29
Stars 3
8
Stars 2
0
Stars 1
3
79ffd31b03954ddf0e755fdfa8d24072

Includes

7 hours on-demand video
Certificate of Completion

About

AulaGEO is a Ge-engineering specialized academy.

Social Network

Udemy Black Friday Sale! Top Courses From $9.99