[WIP] update type inference for string columns

[Moh-Yakoub]

Overview

Closes #249

I've noticed that we have a type deduction error in case the field had NaNs as described in the attached issue.
I furthur noticed that we don't check if the string columns can be casted to a float/int type. I've added extra check to see id a string column can be casted to an int/float column and deduce the proper data_type accordingly.

Update (04/10/2021)

• Adding warning message when NaN columns displayed as histograms
• Resolve NaN series causing error in histogram execute_binning
• Rewrote is_numeric_nan_column to run 2x faster
  

Changes

I've added a logic to

1. Check if the column can be cast to int/double
2. apply the respective data_type inference.

Example Output

The result shows that the two columns mentioned in the issue: # Instances and # Attributes have a correct data_type now

[codecov]

Codecov Report

Merging #343 (eeee236) into master (1a72332) will increase coverage by 0.27%.
The diff coverage is 92.68%.

❗ Current head eeee236 differs from pull request most recent head 02a6a99. Consider uploading reports for the commit 02a6a99 to get more accurate results

@@            Coverage Diff             @@
##           master     #343      +/-   ##
==========================================
+ Coverage   79.98%   80.25%   +0.27%     
==========================================
  Files          50       50              
  Lines        3612     3632      +20     
==========================================
+ Hits         2889     2915      +26     
+ Misses        723      717       -6     

Impacted Files	Coverage Δ
lux/vislib/altair/Choropleth.py	94.20% <0.00%> (ø)
lux/vislib/matplotlib/ScatterChart.py	76.47% <50.00%> (+1.17%)	⬆️
lux/action/univariate.py	90.90% <100.00%> (+0.52%)	⬆️
lux/executor/PandasExecutor.py	96.07% <100.00%> (+0.06%)	⬆️
lux/utils/utils.py	90.12% <100.00%> (+1.23%)	⬆️
lux/vislib/matplotlib/MatplotlibRenderer.py	87.30% <100.00%> (+0.63%)	⬆️
lux/interestingness/interestingness.py	87.56% <0.00%> (+1.08%)	⬆️
lux/vislib/matplotlib/Heatmap.py	98.33% <0.00%> (+1.66%)	⬆️
lux/vislib/altair/ScatterChart.py	96.96% <0.00%> (+3.03%)	⬆️
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1a72332...02a6a99. Read the comment docs.

[dorisjlee]

Thanks @Moh-Yakoub! I made some changes to resolve the related issue in #249 and rewrote the helper function in a more optimized way. Congrats on your first contribution to Lux!