#056 regression

Implementing linear regression, logistic regression and multi-nominal logistic regression with tensorflow-js and node.

Notes

Library Installation/Setup

Uses the following npm modules:

• lodash
• shuffle-seed
• tfjs-node
• memoize
• mnist-data
• node-remote-plot

install with:

npm install

Data and Common Code

• cars.csv
• load-csv.js

Linear Regression

Sources:

• index.js
• linear-regression.js

Run:

$ cd linear-regression
$ node index.js

r2: 0.8059265340599994
Tensor
     [[18.0580692],]

Results:

Logistic Regression

Sources:

• index.js
• linear-regression.js

Run:

$ cd logistic-regression
$ node index.js

0.88

Results:

Multi-nominal Logistic Regression

Sources:

• efficiency-prediction.js
• image-recognition.js
• logistic-regression.js
• memory.js

Run:

$ cd multinominal-logistic-regression
$ node efficiency-prediction.js

0.78

Results:

Credits and References

• Machine Learning with Javascript