hse-aml / competitive-data-science Goto Github PK

Please provide details!

sales_train.csv.gz items.csv item_categories.csv shops.csv

Seems the data is in this project, but I can't get it (perhaps b/c not 'invited')

https://www.kaggle.com/c/competitive-data-science-final-project

Hi,
has someone got this running on Google Colab?

There should be two things we need: the grader package and the data files.
-Anton

There is a bug in week 4 programming assignment notebook where you generate lag features
'After creating a grid, we can calculate some features. We will use lags from [1, 2, 3, 4, 5, 12] months ago.'

The lag features are correct for only target_lag_{} (target_lag_1,2,3 ...) and incorrect for any other lag features.

I documented that bug and the fix in here. Fixing this bug helps me boost my score in LB tremendously.
https://gist.github.com/anhquan0412/330494b051f74eacad3917f43e3ba43a

The 3 EDA files:
EDA_video2.ipynb
EDA_video3_screencast.ipynb
EDA_Springleaf_screencast.ipynb

refer to data files which cannot be found. I like to run my own notebooks locally to try different parameters.

I think there is an issue in the sanity check

test_knn_feats = NNF.predict(X_test[:50])

print ('Deviation from ground thruth features: %f' % np.abs(test_knn_feats - true_knn_feats_first50[44:45]).sum())

Shouldn't it be:

test_knn_feats = NNF.predict(X_test[44:45])

print ('Deviation from ground thruth features: %f' % np.abs(test_knn_feats - true_knn_feats_first50[44:45]).sum())

Because else we are comparing the wrong rows.

There is a line in https://github.com/zyunsg/Advanced-Machine-Learning/blob/master/course2/week1/assignments/PandasBasics.ipynb:

DATA_FOLDER = '../readonly/final_project_data/'

Is the folder missing in the root?

It cannot be checked out under windows.

In compute_KNN_features at the last cell of the notebook the grader is not instantiated consequently cannot submit.

This works:

Hi,

in compute_KNN_features, honours assignment week 4, inside get_features_for_one, it says:

"2. Same label streak: the largest number N, such that N nearest neighbours have the same label."

I find the task label to be very misleading. Literally it means to check the max number of neighbours within the array with the same label. I would reformulate the task label as:

"2. Same label streak: the largest number N, such that the first N nearest neighbours have the same label."

I hope you can understand my point.

Thanks for the great course!

Alessandro

Hi, friends.
I have not found materials for CatBoost basics assignment. Where could I find it?

Whenever I try submitting the assignement I get the error: TypeError: argument of type 'NoneType' is not iterable in line 60 of grader.py

I think the code in line 58 should be if response.status_code == 201: and not if request.status_code == 201:

Please check this Iam unable to submit my assignment due to this issue.
Attaching a screenshot.

Hi. The instruction of the final task of PandasBasics says -

What was the variance of the number of sold items per day sequence for the shop with shop_id = 25 in December, 2014? Do not count the items, that were sold but returned back later.

So the code that followed for getting the rows filtered according to the given condition is -

transactions[(transactions.shop_id==25) & (transactions.date.dt.year==2014) & \
                              (transactions.date.dt.month==12) & (transactions.item_cnt_day>0.0)]

I included (transactions.item_cnt_day>0.0) to only consider items with no returns. And then take the variance accordingly. But this does not seem to work. When I omit the condition I get the answer to be correct. Could you please tell me where I am getting this wrong?