I've been working with this Kaggle dataset and intend to use it to make some time-series predictions from it. I've preprocessed it to the point where now I need to start making some decisions about how best to feed it into the neural-net. Coming from a background in Python's Numpy, I was taught that computations on rows are much more efficient that going wide and doing computations column-wise which is how the kernel would need to read this data if I am understanding its structure correctly. (It's almost 2 years worth of daily data.) The documentation on the subject is not exactly clear.
I just wanted to check if that assumption is correct? And if I should be trying to find a way to transpose this dataset what's the best way to do that? Or if I'm mistaken and using the data how it came out of the pre-processing step will be fine? A sample of the data follows:
<|"title" -> "1984-(roman)", "2015-07-01" -> 421, "2015-07-02" -> 438,
"2015-07-03" -> 351, "2015-07-04" -> 259, "2015-07-05" -> 329,
"2015-07-06" -> 383, "2015-07-07" -> 361, "2015-07-08" -> 333,
"2015-07-09" -> 327|>, <|"title" -> "24-Heures-du-Mans",
"2015-07-01" -> 203, "2015-07-02" -> 188, "2015-07-03" -> 208,
"2015-07-04" -> 169, "2015-07-05" -> 170, "2015-07-06" -> 172,
"2015-07-07" -> 147, "2015-07-08" -> 194,
"2015-07-09" -> 143|>, <|"title" -> "24-Heures-du-Mans-2016",
"2015-07-01" -> 19, "2015-07-02" -> 14, "2015-07-03" -> 20,
"2015-07-04" -> 8, "2015-07-05" -> 10, "2015-07-06" -> 26,
"2015-07-07" -> 24, "2015-07-08" -> 17,
"2015-07-09" -> 9|>, <|"title" -> "2-Broke-Girls",
"2015-07-01" -> 250, "2015-07-02" -> 200, "2015-07-03" -> 179,
"2015-07-04" -> 183, "2015-07-05" -> 204, "2015-07-06" -> 204,
"2015-07-07" -> 212, "2015-07-08" -> 212, "2015-07-09" -> 185|>
