!wget --no-check-certificate 'https://www.dropbox.com/s/5iuef7c9ljj84t6/train_transaction.csv?dl=0' -O train_transaction.csv
!wget --no-check-certificate 'https://www.dropbox.com/s/cmy01z5fw7ohlmd/train_identity.csv?dl=0' -O train_identity.csv
!wget --no-check-certificate 'https://www.dropbox.com/s/7thqkuxnwsa7njj/test_transaction.csv?dl=0' -O test_transaction.csv
!wget --no-check-certificate 'https://www.dropbox.com/s/b40nvbb9e2usd5w/test_identity.csv?dl=0' -O test_identity.csv
!wget --no-check-certificate 'https://www.dropbox.com/s/arkyoz0bel8z4d2/sample_submission.csv?dl=0' -O sample_submission.csv

!pip install catboost -q

def reduce_mem_usage(df):
    NAlist = [] # Keeps track of columns that have missing values filled in.
    for col in tqdm(df.columns):
        if df[col].dtype != object:  # Exclude strings

            # make variables for Int, max and min
            IsInt = False
            col_max_value = df[col].max()
            col_min_value = df[col].min()

            # Integer does not support NA, therefore, NA needs to be filled
            if not np.isfinite(df[col]).all():
                NAlist.append(col)
                df[col].fillna(col_min_value - 1, inplace=True)

            # test if column can be converted to an integer
            col_as_int = df[col].fillna(0).astype(np.int64)
            diff = (df[col] - col_as_int)
            diff = diff.sum()
            if np.abs(diff) < 0.01:
                IsInt = True

            # Make Integer/unsigned Integer datatypes
            if pd.api.types.is_integer_dtype(df[col]):
                if col_min_value >= 0:
                    if col_max_value < 255:
                        df[col] = df[col].astype(np.uint8)
                    elif col_max_value < 65535:
                        df[col] = df[col].astype(np.uint16)
                    elif col_max_value < 4294967295:
                        df[col] = df[col].astype(np.uint32)
                    else:
                        df[col] = df[col].astype(np.uint64)
                else:
                    if col_min_value > np.iinfo(np.int8).min and col_max_value < np.iinfo(np.int8).max:
                        df[col] = df[col].astype(np.int8)
                    elif col_min_value > np.iinfo(np.int16).min and col_max_value < np.iinfo(np.int16).max:
                        df[col] = df[col].astype(np.int16)
                    elif col_min_value > np.iinfo(np.int32).min and col_max_value < np.iinfo(np.int32).max:
                        df[col] = df[col].astype(np.int32)
                    elif col_min_value > np.iinfo(np.int64).min and col_max_value < np.iinfo(np.int64).max:
                        df[col] = df[col].astype(np.int64)

            # Make float datatypes 32 bit
            else:
                df[col] = df[col].astype(np.float32)

    return df, NAlist

INPUT_DIR = '.'

train_transaction = pd.read_csv(os.path.join(INPUT_DIR, 'train_transaction.csv'))
train_identity = pd.read_csv(os.path.join(INPUT_DIR, 'train_identity.csv'))
test_transaction = pd.read_csv(os.path.join(INPUT_DIR, 'test_transaction.csv'))
test_identity = pd.read_csv(os.path.join(INPUT_DIR, 'test_identity.csv'))
sample_submission = pd.read_csv(os.path.join(INPUT_DIR, 'sample_submission.csv'))

df_train = train_transaction.merge(train_identity, how='left', on='TransactionID')
del train_transaction, train_identity
df_train, df_train_NAlist = reduce_mem_usage(df_train)

df_test = test_transaction.merge(test_identity, how='left', on='TransactionID')
del test_transaction, test_identity
df_test, df_test_NAlist = reduce_mem_usage(df_test)