snorkelflow.client.analyses.plot_scatterplot

snorkelflow.client.analyses.plot_scatterplot(node, df, x_axis_field, y_axis_field, label_field, drop_unknown=True, display_legend=True, scatter_kwargs=None, axes_kwargs=None, filter_flag_column=None, color_map='rainbow')

Plot the scatterplot of datapoints across two fields.

To find promising LFs involving a combination of numeric features, the plot_scatterplot() function can be used instead to visualize datapoints. Running %matplotlib notebook in your notebook prior to running this function will open up the plot in interactive mode, allowing for easier exploration of the data.

If you find a view while exploring a scatterplot in interactive mode that would make a promising bounding rectangle LF, then you can use the add_scatterpot_lf function to do so easily. Using the function in the next cell will automatically generate a LF from your current view of the scatterplot. For example, we can examine network data and easily creating a bounding LF for Youtube traffic as follows:

Examples

%matplotlib notebook
df = sf.get_dataset(node, split='dev')
axis_obj = sf.plot_scatterplot(node, df, 'Fwd.Packet.Length.Max', 'Avg.Fwd.Segment.Size', "ProtocolName")

# Run this cell after the view in the interactive scatterplot contains mostly YOUTUBE labels
sf.add_scatterplot_lf(node, axis_obj, 'YOUTUBE')

Parameters

Parameters

Name	Type	Default	Info
node	int		UID of the node.
df	DataFrame		The data frame with a numeric field over which to plot the scatterplot.
x_axis_field	str		Name of the field in the df that contains numeric values to plot on the x axis.
y_axis_field	str		Name of the field in the df that contains numeric values to plot on the y axis.
label_field	str		Name of the field containing labels for plotting on a per-class basis.
drop_unknown	bool	True	If True, remove examples with unknown GT when plotting on a per-class basis.
display_legend	bool	True	If True display a legend of the plotted classes.
scatter_kwargs	Optional[Dict[str, Any]]	None	Arguments to pass through to matplotlib.ax.scatter() (https://matplotlib.org/api/_as_gen/matplotlib.axes.Axes.scatter.html).
axes_kwargs	Optional[Dict[str, Any]]	None	Arguments to pass through to matplotlib.pyplot.gca().set() (https://matplotlib.org/api/_as_gen/matplotlib.figure.Figure.html#matplotlib.figure.Figure.gca) Commonly used axes_kwargs: title: Add a title to the plot xlabel: Edit x-axis label (default is x_axis_field value) ylabel: Edit y-axis label (default is y_axis_field value)
filter_flag_column	Optional[str]	None	Name of column describing whether datapoints have been filtered out are not (can be accessed via show_filtered_flag=True in task.dataset()).
color_map	str	'rainbow'	Name of the matplotlib color map to use when coloring classes (https://matplotlib.org/3.1.0/tutorials/colors/colormaps.html).

Raises

Raises

• ValueError – If x_axis_field is not one of the columns in the given DataFrame

• ValueError – If y_axis_field is not one of the columns in the given DataFrame

• ValueError – If label_field is not one of the coluns in the given DataFrame

• ValueError – If filter_flag_column is not one of the coluns in the given DataFrame

Returns

Returns

An axis object containing information about the plot that can be passed into add_scatterplot_lf to quickly generate a bounding rect LF

Return type

Return type

matplotlib.pyplot.axis