Improve test coverage

[ablaom]

This is currently at 9%. I would say that until coverage is above 70%, adding tests should take priority over adding functionality.

[aa25desh]

Unit Tests

• [x] dtw_distance - equal length of time series

• [x] select_sort

• [x] InvFeatureGen

Integration tests

• [x] TSF

• [x] KNN

[aa25desh]

Remaining

• [ ] dtw_distance - handling unequal length of time series.

There are some extra functions in utils.jl which are not in use as of now.

• [ ] TSdataset - Used to covert the ts files to csv files.

@ablaom for Integration tests I have added check Accuracy > value or should we use predict with random number to get reoccurring output?

Do we need tests for data loading functions?

• [ ] load_dataset

• [ ] load_ts_file

• [ ] ts_dataset

• [ ] load_from_tsfile_to_NDArray

• [ ] load_NDdataset

[ablaom]

Looking good. 60% is great boost 🎉

@ablaom for Integration tests I have added check Accuracy > value or should we use predict with random number to get reoccurring output?

The best tests might be a very simple example where you can work the answer out by hand and check for approximate equality (isapprox or ≈). But this can often be difficult. That's why it's good to break down your algorithm into lots of smaller functions that you can do this for. So, I think Accuracy > value is okay, where value could be some baseline model prediction (eg, constant, feature-independent, prediction).

You can use the codecov tool to find where coverage is low, then decide which of those areas should be prioritized. Follow the links from the codecov .

[ablaom]

Do we need tests for data loading functions?

Couldn't hurt. Start with those with the most complex logic.

[ablaom]

• [x] select_sort

• [x] InvFeatureGen

Where are these? I could not find them.