Adding Legends in Pandas Plot - python

I am plotting Density Graphs using Pandas Plot. But I am not able to add appropriate legends for each of the graphs. My code and result is as as below:-
for i in tickers:
df = pd.DataFrame(dic_2[i])
mean=np.average(dic_2[i])
std=np.std(dic_2[i])
maximum=np.max(dic_2[i])
minimum=np.min(dic_2[i])
df1=pd.DataFrame(np.random.normal(loc=mean,scale=std,size=len(dic_2[i])))
ax=df.plot(kind='density', title='Returns Density Plot for '+ str(i),colormap='Reds_r')
df1.plot(ax=ax,kind='density',colormap='Blues_r')
You can see in the pic, top right side box, the legends are coming as 0. How do I add something meaningful over there?
print(df.head())
0
0 -0.019043
1 -0.0212065
2 0.0060413
3 0.0229895
4 -0.0189266

I think you may want to restructure the way you've created the graph. An easy way to do this is to create the ax before plotting:
# sample data
df = pd.DataFrame()
df['returns_a'] = [x for x in np.random.randn(100)]
df['returns_b'] = [x for x in np.random.randn(100)]
print(df.head())
returns_a returns_b
0 1.110042 -0.111122
1 -0.045298 -0.140299
2 -0.394844 1.011648
3 0.296254 -0.027588
4 0.603935 1.382290
fig, ax = plt.subplots()
I then created the dataframe using the parameters specified in your variables:
mean=np.average(df.returns_a)
std=np.std(df.returns_a)
maximum=np.max(df.returns_a)
minimum=np.min(df.returns_a)
pd.DataFrame(np.random.normal(loc=mean,scale=std,size=len(df.returns_a))).rename(columns={0: 'std_normal'}).plot(kind='density',colormap='Blues_r', ax=ax)
df.plot('returns_a', kind='density', ax=ax)
This second dataframe you're working with is created by default with column 0. You'll need to rename this.

I figured out a simpler way to do this. Just add column names to the dataframes.
for i in tickers:
df = pd.DataFrame(dic_2[i],columns=['Empirical PDF'])
print(df.head())
mean=np.average(dic_2[i])
std=np.std(dic_2[i])
maximum=np.max(dic_2[i])
minimum=np.min(dic_2[i])
df1=pd.DataFrame(np.random.normal(loc=mean,scale=std,size=len(dic_2[i])),columns=['Normal PDF'])
ax=df.plot(kind='density', title='Returns Density Plot for '+ str(i),colormap='Reds_r')
df1.plot(ax=ax,kind='density',colormap='Blues_r')

Related

Couldn't align X axis values with bars on top of them using seaborn barplot with hue [duplicate]

My graph is ending up looking like this:
I took the original titanic dataset and sliced some columns and created a new dataframe via the following code.
Cabin_group = titanic[['Fare', 'Cabin', 'Survived']] #selecting certain columns from dataframe
Cabin_group.Cabin = Cabin_group.Cabin.str[0] #cleaning the Cabin column
Cabin_group = Cabin_group.groupby('Cabin', as_index =False).Survived.mean()
Cabin_group.drop([6,7], inplace = True) #drop Cabin G and T as instances are too low
Cabin_group['Status']= ('Poor', 'Rich', 'Rich', 'Medium', 'Medium', 'Poor') #giving each Cabin a status value.
So my new dataframe `Cabin_group' ends up looking like this:
Cabin Survived Status
0 A 0.454545 Poor
1 B 0.676923 Rich
2 C 0.574468 Rich
3 D 0.652174 Medium
4 E 0.682927 Medium
5 F 0.523810 Poor
Here is how I tried to plot the dataframe
fig = plt.subplots(1,1, figsize = (10,4))
sns.barplot(x ='Cabin', y='Survived', hue ='Status', data = Cabin_group )
plt.show()
So a couple of things are off with this graph;
First we have the bars A, D, E and F shifted away from their respective x-axis labels. Secondly, the bars itself seem to appear thinner/skinnier than my usual barplots.
Not sure how to shift the bars to their proper place, as well as how to control the width of the bars.
Thank you.
This can be achieved by doing dodge = False. It is handled in the new version of seaborn.
The bar are not aligned since it expects 3 bars for each x (1 for each distinct value of Status) and only one is provided. I think one of the solution is to map a color to the Status. As far as i know it is not possible to do thaht easily. However, here is an example of how to do that. I'm not sure about that since it seems complicated to simply map a color to a category (and the legend is not displayed).
# Creating a color mapping
Cabin_group['Color'] = Series(pd.factorize(Cabin_group['Status'])[0]).map(
lambda x: sns.color_palette()[x])
g = sns.barplot(x ='Cabin', y='Survived', data=Cabin_group, palette=Cabin_group['Color'])
When I see how simple it is in R ... But infortunately the ggplot implementation in Python does not allow to plot a geom_bar with stat = 'identity'.
library(tidyverse)
Cabin_group %>% ggplot() +
geom_bar(aes(x = Cabin, y= Survived, fill = Status),
stat = 'identity')

how to make stacked plots for dataframe with multiple index in python?

I have trade export data which is collected weekly. I intend to make stacked bar plot with matplotlib but I have little difficulties managing pandas dataframe with multiple indexes. I looked into this post but not able to get what I am expecting. Can anyone suggest a possible way of doing this in python? Seems I made the wrong data aggregation and I think I might use for loop to iterate year then make a stacked bar plot on a weekly base. Does anyone know how to make this easier in matplotlib? any idea?
reproducible data and my attempt
import pandas as pd
import matplotlib.pyplot as plt
# load the data
url = 'https://gist.githubusercontent.com/adamFlyn/0eb9d60374c8a0c17449eef4583705d7/raw/edea1777466284f2958ffac6cafb86683e08a65e/mydata.csv'
df = pd.read_csv(url, parse_dates=['weekly'])
df.drop('Unnamed: 0', axis=1, inplace=True)
nn = df.set_index(['year','week'])
nn.drop("weekly", axis=1, inplace=True)
f, a = plt.subplots(3,1)
nn.xs('2018').plot(kind='bar',ax=a[0])
nn.xs('2019').plot(kind='bar',ax=a[1])
nn.xs('2020').plot(kind='bar',ax=a[2])
plt.show()
plt.close()
this attempt didn't work for me. instead of explicitly selecting years like 2018, 2019, ..., is there any more efficient to make stacked bar plots for dataframe with multiple indexes? Any thoughts?
desired output
this is the desired stacked bar plot for year of 2018 as an example
how should I get my desired stacked bar plot? Any better ideas?
Try this:
nn.groupby(level=0).plot.bar(stacked=True)
or to prevent year as tuple in x axis:
for n, g in nn.groupby(level=0):
g.loc[n].plot.bar(stacked=True)
Update per request in comments
for n, g in nn.groupby(level=0):
ax = g.loc[n].plot.bar(stacked=True, title=f'{n} Year', figsize=(8,5))
ax.legend(loc='lower center')
Change layout position
fig, ax = plt.subplots(1,3)
axi = iter(ax)
for n, g in nn.groupby(level=0):
axs = next(axi)
g.loc[n].plot.bar(stacked=True, title=f'{n}', figsize=(15,8), ax=axs)
axs.legend(loc='lower center')
Try using loc instead of xs:
f, a = plt.subplots(3,1)
for x, ax in zip(nn.index.unique('year'),a.ravel()):
nn.loc[x].plot.bar(stacked=True, ax=ax)

Pandas groupby results on the same plot

I am dealing with the following data frame (only for illustration, actual df is quite large):
seq x1 y1
0 2 0.7725 0.2105
1 2 0.8098 0.3456
2 2 0.7457 0.5436
3 2 0.4168 0.7610
4 2 0.3181 0.8790
5 3 0.2092 0.5498
6 3 0.0591 0.6357
7 5 0.9937 0.5364
8 5 0.3756 0.7635
9 5 0.1661 0.8364
Trying to plot multiple line graph for the above coordinates (x as "x1 against y as "y1").
Rows with the same "seq" is one path, and has to be plotted as one separate line, like all the x, y coordinates corresponding the seq = 2 belongs to one line, and so on.
I am able to plot them, but on a separate graphs, I want all the lines on the same graph, Using subplots, but not getting it right.
import matplotlib as mpl
import matplotlib.pyplot as plt
%matplotlib notebook
df.groupby("seq").plot(kind = "line", x = "x1", y = "y1")
This creates 100's of graphs (which is equal to the number of unique seq). Suggest me a way to obtain all the lines on the same graph.
**UPDATE*
To resolve the above problem, I implemented the following code:
fig, ax = plt.subplots(figsize=(12,8))
df.groupby('seq').plot(kind='line', x = "x1", y = "y1", ax = ax)
plt.title("abc")
plt.show()
Now, I want a way to plot the lines with specific colors. I am clustering path from seq = 2 and 5 in cluster 1; and path from seq = 3 in another cluster.
So, there are two lines under cluster 1 which I want in red and 1 line under cluster 2 which can be green.
How should I proceed with this?
You need to init axis before plot like in this example
import pandas as pd
import matplotlib.pylab as plt
import numpy as np
# random df
df = pd.DataFrame(np.random.randint(0,10,size=(25, 3)), columns=['ProjID','Xcoord','Ycoord'])
# plot groupby results on the same canvas
fig, ax = plt.subplots(figsize=(8,6))
df.groupby('ProjID').plot(kind='line', x = "Xcoord", y = "Ycoord", ax=ax)
plt.show()
Consider the dataframe df
df = pd.DataFrame(dict(
ProjID=np.repeat(range(10), 10),
Xcoord=np.random.rand(100),
Ycoord=np.random.rand(100),
))
Then we create abstract art like this
df.set_index('Xcoord').groupby('ProjID').Ycoord.plot()
Another way:
for k,g in df.groupby('ProjID'):
plt.plot(g['Xcoord'],g['Ycoord'])
plt.show()
Here is a working example including the ability to adjust legend names.
grp = df.groupby('groupCol')
legendNames = grp.apply(lambda x: x.name) #Get group names using the name attribute.
#legendNames = list(grp.groups.keys()) #Alternative way to get group names. Someone else might be able to speak on speed. This might iterate through the grouper and find keys which could be slower? Not sure
plots = grp.plot('x1','y1',legend=True, ax=ax)
for txt, name in zip(ax.legend_.texts, legendNames):
txt.set_text(name)
Explanation:
Legend values get stored in the parameter ax.legend_ which in turn contains a list of Text() objects, with one item per group, where Text class is found within the matplotlib.text api. To set the text object values, you can use the setter method set_text(self, s).
As a side note, the Text class has a number of set_X() methods that allow you to change the font sizes, fonts, colors, etc. I haven't used those, so I don't know for sure they work, but can't see why not.
based on Serenity's anwser, i make the legend better.
import pandas as pd
import matplotlib.pylab as plt
import numpy as np
# random df
df = pd.DataFrame(np.random.randint(0,10,size=(25, 3)), columns=['ProjID','Xcoord','Ycoord'])
# plot groupby results on the same canvas
grouped = df.groupby('ProjID')
fig, ax = plt.subplots(figsize=(8,6))
grouped.plot(kind='line', x = "Xcoord", y = "Ycoord", ax=ax)
ax.legend(labels=grouped.groups.keys()) ## better legend
plt.show()
and you can also do it like:
grouped = df.groupby('ProjID')
fig, ax = plt.subplots(figsize=(8,6))
g_plot = lambda x:x.plot(x = "Xcoord", y = "Ycoord", ax=ax, label=x.name)
grouped.apply(g_plot)
plt.show()
and it looks like:

Parsing CSV file using Panda

I have been using matplotlib for quite some time now and it is great however, I want to switch to panda and my first attempt at it didn't go so well.
My data set looks like this:
sam,123,184,2.6,543
winter,124,284,2.6,541
summer,178,384,2.6,542
summer,165,484,2.6,544
winter,178,584,2.6,545
sam,112,684,2.6,546
zack,145,784,2.6,547
mike,110,984,2.6,548
etc.....
I want first to search the csv for anything with the name mike and create it own list. Now with this list I want to be able to do some math for example add sam[3] + winter[4] or sam[1]/10. The last part would be to plot it columns against each other.
Going through this page
http://pandas.pydata.org/pandas-docs/stable/io.html#io-read-csv-table
The only thing I see is if I have a column header, however, I don't have any headers. I only know the position in a row of the values I want.
So my question is:
How do I create a bunch of list for each row (sam, winter, summer)
Is this method efficient if my csv has millions of data point?
Could I use matplotlib plotting to plot pandas dataframe?
ie :
fig1 = plt.figure(figsize= (10,10))
ax = fig1.add_subplot(211)
ax.plot(mike[1], winter[3], label='Mike vs Winter speed', color = 'red')
You can read a csv without headers:
data=pd.read_csv(filepath, header=None)
Columns will be numbered starting from 0.
Selecting and filtering:
all_summers = data[data[0]=='summer']
If you want to do some operations grouping by the first column, it will look like this:
data.groupby(0).sum()
data.groupby(0).count()
...
Selecting a row after grouping:
sums = data.groupby(0).sum()
sums.loc['sam']
Plotting example:
sums.plot()
import matplotlib.pyplot as plt
plt.show()
For more details about plotting, see: http://pandas.pydata.org/pandas-docs/version/0.18.1/visualization.html
df = pd.read_csv(filepath, header=None)
mike = df[df[0]=='mike'].values.tolist()
winter = df[df[0]=='winter'].values.tolist()
Then you can plot those list as you wanted to above
fig1 = plt.figure(figsize= (10,10))
ax = fig1.add_subplot(211)
ax.plot(mike, winter, label='Mike vs Winter speed', color = 'red')

Python: Legend has wrong colors on Pandas MultiIndex plot

I'm trying to plot data from 2 seperate MultiIndex, with the same data as levels in each.
Currently, this is generating two seperate plots and I'm unable to customise the legend by appending some string to individualise each line on the graph. Any help would be appreciated!
Here is the method so far:
def plot_lead_trail_res(df_ante, df_post, symbols=[]):
if len(symbols) < 1:
print "Try again with a symbol list. (Time constraints)"
else:
df_ante = df_ante.loc[symbols]
df_post = df_post.loc[symbols]
ante_leg = [str(x)+'_ex-ante' for x in df_ante.index.levels[0]]
post_leg = [str(x)+'_ex-post' for x in df_post.index.levels[0]]
print "ante_leg", ante_leg
ax = df_ante.unstack(0).plot(x='SHIFT', y='MUTUAL_INFORMATION', legend=ante_leg)
ax = df_post.unstack(0).plot(x='SHIFT', y='MUTUAL_INFORMATION', legend=post_leg)
ax.set_xlabel('Time-shift of sentiment data (days) with financial data')
ax.set_ylabel('Mutual Information')
Using this function call:
sentisignal.plot_lead_trail_res(data_nasdaq_top_100_preprocessed_mi_res, data_nasdaq_top_100_preprocessed_mi_res_validate, ['AAL', 'AAPL'])
I obtain the following figure:
Current plots
Ideally, both sets of lines would be on the same graph with the same axes!
Update 2 [Concatenation Solution]
I've solved the issues of plotting from multiple frames using concatenation, however the legend does not match the line colors on the graph.
There are not specific calls to legend and the label parameter in plot() has not been used.
Code:
df_ante = data_nasdaq_top_100_preprocessed_mi_res
df_post = data_nasdaq_top_100_preprocessed_mi_res_validate
symbols = ['AAL', 'AAPL']
df_ante = df_ante.loc[symbols]
df_post = df_post.loc[symbols]
df_ante.index.set_levels([[str(x)+'_ex-ante' for x in df_ante.index.levels[0]],df_ante.index.levels[1]], inplace=True)
df_post.index.set_levels([[str(x)+'_ex-post' for x in df_post.index.levels[0]],df_post.index.levels[1]], inplace=True)
df_merge = pd.concat([df_ante, df_post])
df_merge['SHIFT'] = abs(df_merge['SHIFT'])
df_merge.unstack(0).plot(x='SHIFT', y='MUTUAL_INFORMATION')
Image:
MultiIndex Plot Image
I think, with
ax = df_ante.unstack(0).plot(x='SHIFT', y='MUTUAL_INFORMATION', legend=ante_leg)
you put the output of the plot() in ax, including the lines, which then get overwritten by the second function call. Am I right, that the lines which were plotted first are missing?
The official procedure would be rather something like
fig = plt.figure(figsize=(5, 5)) # size in inch
ax = fig.add_subplot(111) # if you want only one axes
now you have an axes object in ax, and can take this as input for the next plots.

Categories