CODEWITHSUNDEEP
pythonholoviews
Ivan
Ivan

Reputation: 20101

Drop down menu for column visualization in Holoviews

I have a dataframe with time index and some columns, similar to 

dt, col1, col2, col3
12, 0.9, 0.3, 0.1
13, 0.9, 0.3, 0.2
14, 2.9, 0.4, 0.1
15, 0.1, 0.3, 0.7

and I can create a Holoviews Curve wihtout a problem for each column. Is it possible to generate a graph with a drop down menu selecting between [col1, col2, col3]?

Upvotes: 0

Views: 1008

Answers (1)

philippjfr
philippjfr

Reputation: 4080

HoloViews is primarily set up to plot so called "tidy" or long dataframes easily, so to take full advantage of the power of HoloViews you will want to convert the data to that format. The pd.melt function is very useful to convert "wide" to tidy data. Based on your example you could do something like this

df = pd.DataFrame({'dt': range(10), 'col1': np.random.rand(10),
                   'col2': np.random.rand(10), 'col3': np.random.rand(10)})
tidy_df = pd.melt(df, id_vars='dt', var_name='column', value_name='value')
ds = hv.Dataset(tidy_df)
ds.to(hv.Curve, 'dt', 'value', 'column')

Alternatively you can also manually create a HoloMap manually:

df = pd.DataFrame({'dt': range(10), 'col1': np.random.rand(10),
               'col2': np.random.rand(10), 'col3': np.random.rand(10)})
hv.HoloMap({col: hv.Curve(df, 'dt', col) for col in ['col1', 'col2', 'col3']},
           kdims='Column')

Having said all this, your use case is a very common one and the next release of HoloViews will include ways of making this kind of thing much easier, so look out for a user guide about working with pandas DataFrames once that lands.

Upvotes: 1

Related Questions