It is possible to filter and plot it in a descending from sum X on Altair?

Question

I've tried to adapt a script that was answered here before, to filter the top ten suppliers (proveedor_nombre), this means, the ones that got more money in descending order, but never gets to work.

Since I'm not even sure of what I'm doing wrong, is there any way you guys know I can get the desired outcome

Here's the raw code:

# Monto total asignado por proveedor

proveedores_dinero = alt.Chart(df).mark_circle().encode(
    x = alt.X('proveedor_nombre:N'),
    y = alt.Y('sum(monto_contrato_minimo):Q'),
)

proveedores_dinero

And here the output:

Answer 1

It is possible to apply rank on transform_window to filter top n results. However, the circles of missing proveedores will be ploted on the axis x=0, in this case, it worked to filter y values by the top n+1 value. In the following example, I'm plotting the top 3 proveedor of 6 possible.

# Generate raw data
data = {
    'proveedor_nombre': ['proveedor_1', 'proveedor_2', 'proveedor_4', 'proveedor_1', 'proveedor_2', 'proveedor_3',
                         'proveedor_4', 'proveedor_1', 'proveedor_5', 'proveedor_6', 'proveedor_4', 'proveedor_5'
                        ],
    'monto_contrato_minimo': [389, 24, 80, 253, 111, 54, 345, 117, 9, 546, 678, 99]
}
df = pd.DataFrame.from_dict(data)

# Store value of top n proveedores
data_sum = df.groupby('proveedor_nombre').agg({'monto_contrato_minimo':'sum'}).sort_values(by='monto_contrato_minimo', ascending=False)

# Plot top 3 proveedor_nombre
alt.Chart(df).mark_circle().encode(
    x = alt.X('proveedor_nombre:N', sort='-y', scale=alt.Scale(domain=list(data_sum.index[:3]))),
    y = alt.Y('sum(monto_contrato_minimo):Q'),
).transform_window(
    rank='rank(monto_contrato_minimo)',
    sort=[alt.SortField('sum(monto_contrato_minimo)', order='descending')]
).transform_filter(
    (alt.datum.rank < 4)  
).transform_filter(
    (alt.datum.monto_contrato_minimo > data_sum.iloc[3][0]) 
).properties(
    width=500, height=300
)

will result in