h0rton.trainval_data.data_utils¶
Module Contents¶
Functions¶
whiten_pixels(pixels) |
|
asinh(x) |
|
plus_1_log(linear) |
Add 1 and take the log10 of an image |
rescale_01(unscaled) |
Rescale an image of unknown range to values between 0 and 1 |
whiten_Y_cols(df, mean, std, col_names) |
Whiten (in place) select columns in the given dataframe, i.e. shift and scale then so that they have the desired mean and std |
log_parameterize_Y_cols(df, col_names) |
Whiten (in place) select columns in the given dataframe, i.e. shift and scale then so that they have the desired mean and std |
-
h0rton.trainval_data.data_utils.plus_1_log(linear)[source]¶ Add 1 and take the log10 of an image
linear : torch.Tensor of shape [X_dim, X_dim]
- torch.Tensor
- the image of the same input shape, with values now logged
-
h0rton.trainval_data.data_utils.rescale_01(unscaled)[source]¶ Rescale an image of unknown range to values between 0 and 1
unscaled : torch.Tensor of shape [X_dim, X_dim]
- torch.Tensor
- the image of the same input shape, with values now scaled between 0 and 1
-
h0rton.trainval_data.data_utils.whiten_Y_cols(df, mean, std, col_names)[source]¶ Whiten (in place) select columns in the given dataframe, i.e. shift and scale then so that they have the desired mean and std
df : pd.DataFrame mean : array-like
target mean- std : array-like
- target std
- col_names : list
- names of columns to whiten
-
h0rton.trainval_data.data_utils.log_parameterize_Y_cols(df, col_names)[source]¶ Whiten (in place) select columns in the given dataframe, i.e. shift and scale then so that they have the desired mean and std
df : pd.DataFrame mean : array-like
target mean- std : array-like
- target std
- col_names : list
- names of columns to whiten