How to use this app

This app lets you upload and use custom datasets (or use built-in datasets popular in machine learning), train and test these models using various binary classification or regression algorithms (select 'Algorithms' from the top menu to learn more) and save their results (known as entries). You can then create graphs comparing the train and test accuracy (or R2 score for regression algorithms) of a particular dataset, algorithm and model parameter.

Please Login To Use This App

How to train and test a model for a built-in dataset

The models that you train and test are known as entries. To create an entry:

  1. Select ‘Entries’ from the top menu.
  2. Click/tap ‘Add Entry’.
  3. Under ML type (the type of algorithm), select ‘binary classification’.
  4. Select a dataset (try the Wisconsin Breast Cancer dataset to start).
  5. Select an algorithm (try K-Nearest Neighbours).
  6. Enter a number for the parameter 1 value: K (# of neighbours) (this can be any positive integer value, choose a number between 1 and 10).
  7. Click or tap the ‘submit’ button.

You will then be taken to the ‘Entries’ page, which shows a list of all the models you have trained and tested. For binary classification algorithms, you will see the train/test accuracy results (which is represented by a decimal number from 0 to 1, where 1 equals 100% accuracy). For regression algorithms, you will see the R2 score, which represents how well a model fits a regression dataset (again represented by a decimal number from 0 to 1, where 1 means the model fits the dataset 100%). From here you can either add another entry, edit or delete any entries you’ve made.

How to create a graph to plot the train/test accuracy/R2 score of an algorithm & dataset

  1. Add at least 3 entries for the same dataset, algorithm, and parameter number (which is 1 by default).
  2. Select 'Graphs' from the top menu.
  3. Click/tap ‘Add Graph’.
  4. Choose the ML type, dataset, algorithm, and parameter to plot (1 by default on most algorithms, i.e. enter the number 1).
  5. Click or tap the ‘submit’ button.

You will then be taken to the Graphs page, where you should see the title of the graph you just created. Click on the title of the graph to view the graph.

How to upload a custom dataset
  1. Select ‘Datasets’ from the top menu.
  2. Click/tap 'Datasets'.
  3. Enter and title and description of the dataset.
  4. Under ML Type, select the type of algorithm, e.g. ‘binary classification’.
  5. Click or tap the ‘upload file’ and locate the file on your hard drive.
  6. Select the filetype (formats supported include CSV, Excel .xlsx and .xls)*.
  7. Click or tap the ‘submit’ button.

You will be taken to the ‘Datasets’ page, and you should now see your dataset added to the list. You can then make entries and graphs using this custom data set. Note that if you delete this dataset, any entries or graphs you make using this dataset will be deleted also.

*(for datasets, the last column will be the target or output variable, and by default Excel files will ignore the first row of the dataset as this is header data, usually the title, which is not needed during training/testing and is provided by the user when uploading the dataset. Also note that for user uploaded binary classification datasets, the target values must either be 0 or 1, or 0 or -1)