In [78]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
from IPython.display import Image, display
pd.options.display.max_colwidth = 100
In [2]:
dane_star = pd.read_csv('starwars.csv')
In [3]:
dane_star.head(3)
Out[3]:
In [4]:
dane_star.head(5)
Out[4]:
In [5]:
dane_star.describe()
Out[5]:
In [22]:
dane_star.sort_values('Minifigs',ascending=False).head()
Out[22]:
In [23]:
dane_star.sort_values('Pieces',ascending=False).head()
Out[23]:
In [39]:
dane_star.sort_values('Pieces',ascending=True).head()
Out[39]:
In [14]:
dane_star.loc[dane_star.Minifigs>20,:]
Out[14]:
In [55]:
def get_link(setid):
link = dane_star.loc[dane_star.SetID==setid,'ImageURL'].reset_index(drop=True)[0]
return link
#Image(url=link[0])
In [65]:
Image(url=get_link(5860))
Out[65]:
In [64]:
dane_star['USPrice'].max()
Out[64]:
In [70]:
dane_star.loc[dane_star['USPrice'] == dane_star['USPrice'].max(),:]
Out[70]:
In [73]:
najstarsze = dane_star.loc[dane_star['Year'] == dane_star['Year'].min(),:]
In [74]:
Image(url=get_link(3826))
Out[74]:
In [79]:
for ii, row in najstarsze.iterrows():
display(Image(url=get_link(row.SetID)))
In [80]:
najmlodsze = dane_star.loc[dane_star['Year'] == dane_star['Year'].max(),:]
In [85]:
for ii, row in najmlodsze.iterrows():
try:
display(Image(url=get_link(row.SetID)))
except ValueError:
print("Oops! ")
In [87]:
najmlodsze.head(6)
Out[87]:
In [89]:
najdrozszeEU = dane_star.loc[dane_star['EUPrice'] == dane_star['EUPrice'].max(),:]
najdrozszeEU
Out[89]:
In [94]:
Image(url=get_link(5860))
Out[94]:
In [93]:
najdrozszeUS = dane_star.loc[dane_star['USPrice'] == dane_star['USPrice'].max(),:]
najdrozszeUS
Out[93]:
In [97]:
dane_star['Year'].hist()
Out[97]:
In [103]:
sns.distplot(dane_star['Year'],bins=dane_star['Year'].drop_duplicates().sort_values(),norm_hist=False)
Out[103]:
In [132]:
to_plot_year = dane_star.loc[:,['Year','SetID']].groupby('Year').agg('count').reset_index()
#to_plot_year
In [131]:
g = sns.barplot(y='SetID',x='Year',data=to_plot_year)
plt.xticks(rotation=45)
Out[131]:
In [134]:
stas = dane_star.loc[dane_star['Year'] == 2008,:]
In [135]:
for ii, row in stas.iterrows():
try:
display(Image(url=get_link(row.SetID)))
except ValueError:
print("Oops! ")
In [137]:
ant = dane_star.loc[dane_star['Year'] == 2011,:]
In [139]:
for ii, row in ant.iterrows():
try:
display(Image(url=get_link(row.SetID)))
except ValueError:
print("Oops! ")
In [ ]: