In [171]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from wordcloud import WordCloud
In [45]:
wine_df = pd.read_csv("wine.csv")
wine_df.head() #look at the head of our df
Out[45]:
Brief summary of the Wine dataframe¶
In [40]:
print('Shape\n',wine_df.shape)
print('--------------------------------------------------------------------------')
print('Number of rows\n', len(wine_df))
print('--------------------------------------------------------------------------')
print('Column headers\n', wine_df.columns)
print('--------------------------------------------------------------------------')
print('Data types\n', wine_df.dtypes)
print('--------------------------------------------------------------------------')
print('Index\n', wine_df.index)
print('--------------------------------------------------------------------------')
print('Missing Data\n',wine_df.isnull().sum())
print('--------------------------------------------------------------------------')
print('Duplicated Rows\n', wine_df.duplicated().sum())
print('--------------------------------------------------------------------------')
In [27]:
#droping the extra index column.
wine_df.drop('Unnamed: 0',1, inplace=True)
In [98]:
#delete the rows that have identical values in each columns
wine_df.drop_duplicates(inplace=True)
- THE 5 MOST EXPENSIVE WINES AND THEIR COUNTRY
In [51]:
wine_df.loc[:,['country','designation', 'price']].sort_values('price', ascending=False).head().reset_index()
Out[51]:
Wine Spectator's 100-Point Scale¶
Wine Spectator tasters review wines on the following 100-point scale:
- 95-100 Classic: a great wine
- 90-94 Outstanding: a wine of superior character and style
- 85-89 Very good: a wine with special qualities
- 80-84 Good: a solid, well-made wine
- 75-79 Mediocre: a drinkable wine that may have minor flaws
- 50-74 Not recommended
https://www.winespectator.com/articles/scoring-scale
SO WE'RE GOING TO GROUP THE POINTS COLUMN ACCORDINGLY
In [54]:
wine_df.points.describe()
Out[54]:
In [195]:
wine_df['scoring_scale'] = pd.cut(wine_df.points, (84,89,94,100), labels=['very_good', 'outstanding', 'classic'])
- NOW LET'S SEE HOW THE POINT OR THE SCORING SCALE INFLUENCE THE PRICE OF WINES
In [196]:
wine_df.loc[wine_df.price > 200,['designation','scoring_scale']].head(10)
Out[196]:
- LET'S GROUPE THE DF BY SCORING SCALE AND COUNT TO SEE HOW THE COMPANIES ARE DOING IN TERMS OF PRODUCING QUALITY WINE
In [197]:
print('not too bad')
wine_df.groupby("scoring_scale").designation.count()
Out[197]:
In [151]:
sns.countplot(x='scoring_scale', data=wine_df)
Out[151]:
- IF I HAD TO RECOMEND YOU SOME WINE IN THIS DF BELOW IS THE TOP 5 CLASSIC WINE
In [96]:
wine_df.loc[wine_df['scoring_scale'] == 'classic', ['country','designation', 'price','variety','points']].head()
Out[96]:
- LET'S HAVE QUICK LOOK AT THE PRICE SCALE
In [1]:
a = '''this is a list of over 10,000 Fast Food restaurants provided by Datafiniti's Business Database. The dataset includes the restaurant's address, city, latitude and longitude coordinates, name, and more. Note that this is a sample of a large dataset The dataset contains 8 Columns and 10000 rows, the columns are mostly of Categorical variables except for longitude and latitude columns which are Continuous variables. See that the Dataset did not come with a Land Area Column I have decided to bring in another Dataset that has Land Area data and join it to our original Dataset this will help us to determine which state has the most Fast Food per meter or mile square. Some of the Questions discuss in the analysis include but not limited: - States with the most and least McDonald's per capita - Fast Food restaurants per capita for all states - Fast Food restaurants with the most locations nationally - States with the most and least Fast Food restaurants per capita - Cities with the most Fast Food restaurants per capita - States with the most and least Fast Food restaurants per capita - The number of Fast Food restaurants per capita We have taken full advantage of the Longitude and Latitude data to generate some informative plot geographically. '''
b = a.strip()
print(b)
In [129]:
sns.distplot(wine_df.price.dropna()) # WE KNOW THAT WE HAVE SOME MISSING VALUES IN PRICE COLUMN SO WE DROP THOSE MISSING VALUES FIRST
Out[129]:
TOP 5 WINE MANUFACTURER COUNTRIES¶
In [193]:
top_pro = wine_df.country.value_counts()
top_5_pro = top_pro.iloc[0:5]
top_5_pro.plot(kind='barh', color=['r','b','g','r','y'])
Out[193]:
In [ ]:
## MOST TOP 5 WINE VARIETIES
In [162]:
top_5_variety = wine_df.variety.value_counts().head()
top_5_variety.plot(kind='barh', color=['r', 'w','r','r','r'], edgecolor = 'black')
Out[162]:
In [100]:
wine_df.isnull().sum()
Out[100]:
ALTHOUGH THERE'S A GOOD AMOUNT OF MISSING VALUE, IT DID NOT AFFECT GREATLY OUR EDA.
but if we had to fix the columns with missing values I think K nearest neighbor will appropriate in this case. because, columns like region_1 and region_2 have a very close relationship with the country, province, description and some other coolums. however for the missing value in price column I was thinking of interpolate() method of filling or using the average price of their scoring_scale.
In [ ]:
# LET'S GENERATE SOME WORD CLOUD FOR PROVINCES AND VARIETIES
In [199]:
province = list(wine_df.province) #CREATE A LIST OF PROVINCES
province = ' '.join(province) # JOIN THE PROVINCES TO GET ONE STRING
plt.figure(figsize=(12,18)) #SET THE CANVAS TO SUIT OUR VIZ
wordcloud = WordCloud().generate(province) #CREATE A WORDCLOUD AND LET THE MODULE DECIDE THE WIDTH AND HEIGHT
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.margins(x=0, y=0)
In [190]:
variety = list(wine_df.variety)
variety = ' '.join(variety)
plt.figure(figsize=(12,18))
wordcloud = WordCloud(background_color='white').generate(variety)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.margins(x=0, y=0)
In [ ]: