Linear regression line on a scatter plot in python

Question

I am trying add a linear regression line to my scatter plot, but the way my data are stored and/or structured make the task pretty tricky for me. I managed to create a scatter plot with my data using "sns.scatterplot", but couldn't replicate it with "sns.regplot".

By using the following piece of code, I was able to produce the figure bellow :

sns.set_theme(color_codes=True)

sns.scatterplot(x=100, y=dict_all_mv['dict_100'])
sns.scatterplot(x=90, y=dict_all_mv['dict_90'])
sns.scatterplot(x=80, y=dict_all_mv['dict_80'])
sns.scatterplot(x=60, y=dict_all_mv['dict_60'])
sns.scatterplot(x=40, y=dict_all_mv['dict_40'])
sns.scatterplot(x=20, y=dict_all_mv['dict_20'])

plt.show()

Scatter plot without regression line

The data displayed on the y axis are stored in nested dictionaries. The type and a print of one of the dictionaries are displayed bellow :

print(type(dict_all_mv['dict_100']))
print(dict_all_mv['dict_100'])

<class 'dict'>
{'P1S1': 0.2825781935483871, 'P2S1': 0.1566529827586207, 'P3S1': 0.16534197660818717, 'P4S1': 0.2463573039215686, 'P5S1': 0.21233809917355378, 'P6S1': 0.2780758316831682, 'P7S1': 0.261270830357143, 'P8S1': 0.25920007258064504, 'P9S1': 0.17294190131578946, 'P10S1': 0.18588214503816788, 'P1S2': 0.1870260291970803, 'P2S2': 0.2059823125, 'P3S2': 0.17782266216216217, 'P4S2': 0.1835217697841726, 'P5S2': 0.21022608870967743, 'P6S2': 0.2695895306122449, 'P7S2': 0.2393472735042735, 'P8S2': 0.18402302247191013, 'P9S2': 0.17260260273972605, 'P10S2': 0.20730387826086952}

I want to plot all the values stored in each dictionaries on the y axis, while the x axis values are equals to 100 for each values of 'dict_100', 90 for each values of 'dict_90', 80 for 'dict_80', etc. I have tried some options to plot a linear regression line with sns.regplot and lmplot, but without any success. Here is one of my failed attempts :

sns.regplot(x=100,y=dict_all_mv['dict_100'])
sns.regplot(x=90,y=dict_all_mv['dict_90'])
sns.regplot(x=80,y=dict_all_mv['dict_80'])

plt.show()

I got this error message :

AttributeError: 'int' object has no attribute 'shape'

My goal is to obtain a graph like this one:

Plot with a discrete x variable showing means a linear regression line

I am pretty sure the solution is to use regplot or lmplot based on the descriptions made at seaborn.pydata.org, but I don't know how to structure the code to get the results that I want.

So any help is welcome ! :)

You need to put your data into a so-called "tidy" dataframe. Third sentence from the intro seaborn (emphasis mine): "Its plotting functions operate on dataframes and arrays containing whole datasets and internally perform the necessary semantic mapping and statistical aggregation to produce informative plots" — Paul H
– Paul H, Commented Mar 25, 2022 at 16:21
I wasn't aware about this point, I am going to do some research about tidy Dataframes, thank you for the info ! — Clément Chéry
– Clément Chéry, Commented Mar 25, 2022 at 20:07

JohanC · Accepted Answer · 2022-03-25 17:25:34Z

1

To have the y-values in "long form", you can concatenate the values in the dictionaries:

ys = [yi for di in dict_all_mv.keys() for yi in dict_all_mv[di].values()]

The corresponding x-values can be obtained similarly, but instead of the value from the dictionary, the numeric value at the end of the key is added:

xs = [int(di[5:]) for di in dict_all_mv.keys() for yi in dict_all_mv[di].values()]

With these two lists you can call sns.regplot:

sns.regplot(x=xs, y=ys)

answered Mar 25, 2022 at 17:25

JohanC

81.4k8 gold badges54 silver badges90 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Linear regression line on a scatter plot in python

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related