Daily reports for Machine Learning features in Scilab
Contents
-
Daily reports for Machine Learning features in Scilab
- Introduction
-
Daily reports
- 15th May
- 16th May
- 17th May
- 18th May
- 19th May
- 21st May
- 22nd May
- 23rd May
- 24th May
- 25th May
- 26th May
- 27th May
- 28th May
- 29th May
- 30th May
- 31st May
- 1st June
- 2nd June
- 3rd June
- 4th June
- 5th June
- 6th June
- 7th June
- 8th June
- 9th June
- 10th June
- 11th June
- 12th June
- 13th June
- 14th June
- 15th June
- 16th June
- 17th June
- 18th June
- 19th June
- 20th June
- 21st June
- 22nd June
- 23rd June
- 24th June
- 25th June
- 26th June
- 27th June
- 28th June
- 29th June
- 30th June
- 1st July
- 2nd July
- 3rd July
- 4th July
- 5th July
- 6th July
- 8th July
- 9th - 12th July
- 13th July - 31st July
- 1st August - 6th August
Introduction
Proposal : Machine Learning features in Scilab
An effort by Soumitra Agarwal under the mentors -
- Aashay Singhal
- Mandar Deshpande
Daily reports
15th May
- Implementation of the Jupyter hub setup initiated on the local machine after the yyintegration of last year's implementation.
16th May
- Jupyter hub only supports python3 so the idea was kept on hold.
- Contacted Simon and Clement for further discussion and python3 guidelines.
17th May
- Set up a virtual instance using google cloud.
- Initialisation of the development and experimentation forking procedure for the summer.
18th May
- Installed anaconda (and other required packages) onto the google vm
19th May
- Transfer of the file to the local machine from the cloud instance completed using scp
- Set up a daily update page for the project
- Set up project wiki
Set up github repository
21st May
- Set up automated scripts to start the ipython kernel on the remote host from the client
- After the kernel is started, a copy script can be run to automatically get the json config file on the local machine
22nd May
- Initialise development procedure with linear regression scripts
Add dataset from Kaggle to the github repo.
23rd May
- Formulate addition of a standalone toolbox for machine learning in Scilab
- Formulate threading mechanism to run the scripts from experimentation independently
24th May
- Add polynomial regression to the set of available scripts in the development section
25th May
- Add Kmeans clustering to the development section
- Separate development section into algorithms, pre-processing and visualisation
26th May
- Add Normalisation and train test split to the pre-processing section under the development header
27th May
- Add Scaling to the pre-processing section
- Formulate the authentication mechanism for the experimentation aspect of the project
- Set up GCP machine initiated by mentor
28th May
- Add Naive Bayes to the algorithms section under development (Gaussian)
29th May
- Add decision tree classification to the algorithms section (Training)
- Begin work on the authentication mechanism to prevent a user from doing anything other than initialising a kernel and copying its json
30th May
- Add command to the authorized_keys option for OpenSSH for the GCP server
- Complete decision tree classification with prediction test
31st May
- Update the project wiki
- Send concise report on the mailing list
1st June
- Updated readme with function usage
- Add comments to code for usage
2nd June
- Prepare timeline with agendas to complete to ensure completion of project
3rd June
- Commence process of automation for the experimentation part with the GCP server
- Add mechanism to copy dataset and scripts to the server
4th June
- Figure out starting a kernel on the server using python (nohup) with the same python_local script as earlier
5th June
- Complete automation architecture with copying mechanism to get the attributes file back
- Figure out kernel tagging mechanism to a user of the GCP server
- Start a new kernel everytime a user wants to run a model and discard previous
6th June
- Implement mechanism to generate missing values and add it to the preprocessing macros setup
7th June
- Implement 1 hot encoding mechanism and add it to the preprocessing macros
- Implement binarization mechanism with the preprocessing macros using the earlier implemented normalization
8th June
- Implement principal component analysis for preprocessing
9th June
- Add single layer perceptron to the set of algorithms in the standalone toolbox
- Add K-Nearest Neighbours classification to the algorithms set
10th June
- Add Ridge regression to algorithms toolbox
- Add Lasso regression to algorithms toolbox
11th June
- Add Elastic net to the set of algorithms
- Add spectral clustering to the set of algorithms
12th June
- Add tests for Elastic Net, Kmeans, Spectral Clustering, Lasso regression, KNN classification and decision tree classification
13th June
- Add LARS regression and multinomial Naive Bayes with tests
14th June
- Add Ensemble learning method (with result comparison) for the set of algorithms
15th June
- Add regression tree learning algorithm to the set of given algorithms
16th June
- Added K-Mediods clustering algorithm to algorithms set
- Added affinity propagation to the set of algorithms
17th June
- Add random forest classification algorithm
18th June
- Add support vector regression for set of algorithms
- Add support vector classification for set of algorithms
19th June
- Start neural network initial setup
- Begin reading on building the toolbox
20th June
- Complete neural network training part
- Add oversampling class rebalancing macro
21st June
- Complete neural network prediction
22nd June
- Add class rebalancing using clustering
- Send updates to the mailing list
23rd June
- Complete toolbox build
24th June
- Add demos
- Add readme
- Add tests
25th June
- Resume experimentation. Initiate packet tests
26th June
- Complete tests for packets
27th June
- Initiate script migration to server
28th June
- Complete migration setup
29th June
- Update template files for toolbox
30th June
- Complete url method for server based learning
1st July
- Publish toolbox
2nd July
- Complete authentication mechanism for Experimentation
3rd July
- Complete custom script for Experimentation
4th July
- Complete custom script with URL download
5th July
- Start toolbox structuring for cloud
6th July
- Perform minor corrections ML toolbox
8th July
- Publish a guide to mentors
- Add scripts to server
9th - 12th July
- Complete cloud machine learning toolbox
13th July - 31st July
- Add multiple enhancements to the cloud toolbox (separating URL download, prevent recurrent prompt for username/pass)
- Structure enhancements into toolbox
1st August - 6th August
- Scrub and structure final code
- Complete blog