()
| 14 | |
| 15 | |
| 16 | def getKaggleMNIST(): |
| 17 | # MNIST data: |
| 18 | # column 0 is labels |
| 19 | # column 1-785 is data, with values 0 .. 255 |
| 20 | # total size of CSV: (42000, 784) |
| 21 | train = pd.read_csv('../large_files/train.csv').values.astype(np.float32) |
| 22 | train = shuffle(train) |
| 23 | |
| 24 | Xtrain = train[:-1000,1:] |
| 25 | Ytrain = train[:-1000,0].astype(np.int32) |
| 26 | |
| 27 | Xtest = train[-1000:,1:] |
| 28 | Ytest = train[-1000:,0].astype(np.int32) |
| 29 | |
| 30 | # scale the data |
| 31 | Xtrain /= 255. |
| 32 | Xtest /= 255. |
| 33 | # scaler = StandardScaler() |
| 34 | # Xtrain = scaler.fit_transform(Xtrain) |
| 35 | # Xtest = scaler.transform(Xtest) |
| 36 | |
| 37 | return Xtrain, Ytrain, Xtest, Ytest |
| 38 | |
| 39 | |
| 40 | def get_spiral(): |
no outgoing calls
no test coverage detected