# plotly error: ValueError: The data property of a figure may only be assigned a list or tuple that contains a permutation of a subset of itself

I am simulating some data and trying to plot various samples of it using `plotly` and `ipythonwidgets`. I created dropdowns to let people choose the sample size and the number of samples that they want to collect from a population distribution generated with this:

``````from random import seed
from numpy.random import normal, negative_binomial, binomial
import pandas as pd

population_N = 1000000
population_data = pd.DataFrame({
"data.normal": normal(0, 1, population_N),
"data.poisson": negative_binomial(1, 0.5, population_N),
"data.binomial": binomial(1, 0.5, population_N)
})
``````

For them to sample (either with or without a condition) and to average over the sample for multiple samples, I created the following function:

``````def custom_simple(df, sample_size, type = "random"):
"""
Description
----
Take the population data.frame and generate a sample with a specific sample size

Parameters
----
df(pd.DataFrame): the population dataset
sample_size(int): the number of rows in the sample
type(str): if random, pull a random sample

Returns
----
sample(pd.DataFrame)
"""
if type == "random":
single = df.sample(sample_size)
else:
condition = df["data.normal"] < 1
single = df[condition].sample(sample_size)
return sample

def mean_samples(df, sample_size, type = "random", num = 1):
"""
Description
----
Take each sample and calculate the mean for each variable in the sample

Parameters
----
df(pd.DataFrame): the population dataset
sample_size(int): the number of rows in the dataset
type(str): if random, then do random sample
num(int): number of samples to take

Returns
----
sample_means(pd.DataFrame)
"""
sample_means = pd.DataFrame()
def repeated_sample(df, sample_size, type = type, num = num):
"""
Description:
----
Take the population dataset and come up with a specified number of samples

Parameters
----
df(pd.DataFrame): the population dataset
sample_size(int): the number of rows in the dataset
type(str): if random, then randomly sample
num(int): the number of samples to generate

Returns
----
sample_list(list(pd.DataFrame)): a list of samples stored as DataFrames
"""
sample_list = [custom_simple(df, sample_size, type) for n in range(num)]
return sample_list
raw_samples = repeated_sample(df, sample_size, type = type, num = num)
for i in raw_samples:
sample_mans = pd.concat([sample_means, i.mean(axis = 0).to_frame().T])
return sample_means
``````

What these functions essentially do is take the `population_data` dataframe object, uses `pd.DataFrame.sample()` to pass varying sample sizes and to then repeat these samples to then be averaged over.

I execute this function with a for loop that stores a dictionary element which is the column average for each sample size for each sample number:

``````sample_data = {}
sample_sizes = [20, 50, 100, 200, 500, 1000, 2000]
num_of_samples = [1, 2, 5, 10, 20, 50, 100]

for j in num_of_samples:
for i in sample_sizes:
sample_data['size_{}_sample_{}'.format(i,j)] = mean_samples(df = population_data, sample_size = i, type = "random", num = j)
sample_data["population_data"] = population_data
``````

I then create a function that uses `ipython.widgets.Dropdown` to give me a dropdown for various sample and number of samples which should connect to one of the `dict` objects in `sample_data`. The function also contains information to pass chosen `dict` object to a plotly histogram.

``````def samples_histogram(dict, variable, type = "random"):
sample_widget = widgets.Dropdown(
options = ["20", "50", "100", "200", "500", "1000", "2000"],
value = "100",
description = "Sample size:"
)
sampling_widget = widgets.Dropdown(
options = ["2", "5", "10", "20", "50", "100"],
value = "2",
description = "# of samples"
)
trace = go.Histogram(x = dict.get("population_data")[variable])
fig = go.FigureWidget(data = trace)
def response(change):
if sample_widget.value == "20" and sampling_widget.value == "2":
temp_df = dict.get("size_20_sample_1")
temp_df = list(dict.items())
temp_df = temp_df[temp_df=='size_20_sample_1']
with fig.batch_update():
fig.data = [temp_df[variable].tolist()]
sample_widget.observe(response, names = "value")
sampling_widget.observe(response, names = "value")
box = widgets.VBox([
sample_widget,
sampling_widget,
fig
])
display(box)
``````

The problem I am running into, is that the `dict` object that I am passing, even when I convert it to a list (see this line):

``````        with fig.batch_update():
fig.data = [temp_df[variable].tolist()]
``````

It gives me this error:

``````ValueError: The data property of a figure may only be assigned
a list or tuple that contains a permutation of a subset of itself.
Received element value of type <class 'list'>
``````

I am not sure if there is some way I can transform the objects I am passing to play nicer with plotly or if I am just missing something.