Skip to content
rootpath
creator
rootpath
More
Share
Explore
root

icon picker
Garden

DataFrames

The DataFrame is like a super data structure, holding other data structures inside of it.

Construction

Think of some data that you can put into sequences of containers (a.k.a. rows of columns)
3 x 4 dataset:
1








2








3








There are no rows in this table
⁠
DataFrames are made from rows and columns. The rows and columns have a list-like structure, and can be
constructed with lists. Create an empty DataFrame, and give it a column and a list. Then, create a another, and
add it to the first.
Code DataFrame: sql, my project,
1
garden = root.garden
​
2
quark['type']=[moonflower1, moonflower2]
type
0 moonflower1
1 moonflower2
2
3
quarkFeatures = pd.DataFrame(
columns=['charge', 'bucket'])

quarkFeatures['charge'] = [1,-1]
quarkFeatures['bucket'] = ['fermion', 'fermion']
charge bucket
0 1 fermion
1 -1 fermion
4
quarkFeatures['type']=quark
charge bucket type
0 1.0 fermion up
1 -1.0 fermion down
2 0.5 fermion beauty
There are no rows in this table
⁠

One of the coolest things about DataFrames are that they evolved to work with real data. In fact, you can
give Pandas a data file, and it will construct for you a DataFrame. Let's make a data file, and then load it up.
Load a CSV:
1
quanta = pd.read_csv('~/data/particles.csv')
examples in nature quanta charge type
0 sunlight photon 0.00 boson
1 lightning electron -1.00 fermion
2 matter up quark 0.66 fermion
3 matter down quark -0.33 fermion
4 moonlight photon 0.00 boson
2
quanta = pd.read_csv('~/data/particles.csv')
examples in nature quanta charge type
0 sunlight photon 0.00 boson
1 lightning electron -1.00 fermion
2 matter up quark 0.66 fermion
3 matter down quark -0.33 fermion
4 moonlight photon 0.00 boson
There are no rows in this table
⁠

Editing the DataFrame

Your DataFrame can be grouped, or reshuffled.
1
quantaTypeXCharge =
quanta.groupby(['charge'])[['quanta','type']].max()
quanta type
charge
-1.00 electron fermion
-0.33 down quark fermion
0.00 photon boson
0.66 up quark fermion
There are no rows in this table
⁠
If you need to get only subsets of data, you can slice them.
1
proton.loc[1:2]
proton.loc[1:2]
2
cols = ['down', 'beauty']

print(proton.loc[1:2, cols])
cols = ['down', 'beauty']

print(proton.loc[1:2, cols])
There are no rows in this table
⁠

Plotting Data

Once your data is loaded into the frame, you can see it, by plotting it.
write line, scatter, and box plots:
1
line
cluster = pd.read_csv('~/data/cluster1.csv')
cluster.plot(kind= 'line', x='d', y='sc p')
plt.show()
2
scatter
cluster = pd.read_csv('~/data/cluster1.csv')
cluster.plot(kind= 'scatter', x='d', y='sc p')
3
box
cluster = pd.read_csv('~/data/cluster1.csv')
cluster.plot(kind= 'box', subplots=False, x='d', y=['sc p', 'cc p'])
There are no rows in this table
⁠

Merging DataFrames

Merging DataFrames is accomplished through stacking , joining, and merging
We can stack DataFrames as rows by appending. Append two Series (1-D DataFrames):
1
neutrinos = pd.Series(['electron neutrino',
'muon neutrino',
'tau neutrino'])

leptons = leptons.append(neutrinos)
print(leptons)
0 electron
1 muon
2 tau
0 electron neutrino
1 muon neutrino
2 tau neutrino
There are no rows in this table
⁠
We can stack both DataFrames vertically and horizontally by specifying the axis by concatenating.
1
fermions = pd.read_csv('~/data/fermions.csv')
bosons = pd.read_csv('~/data/bosons.csv')

buckets = [fermions, bosons]
print(pd.concat(buckets, axis=1))

quanta type quanta type
0 electron fermion photon boson
1 up quark fermion NaN NaN
2 down quark fermion NaN NaN
There are no rows in this table
⁠
Stack a multi-indexed DataFrames with concatenation, by adding keys, as well as axis, like so:
1
vertical =
pd.concat(buckets, keys=['fermions','bosons'], axis=0)
print(vertical)

horizontal =
pd.concat(buckets, keys=['fermions','bosons'], axis=0)
print(horizontal)
quanta type
fermions 0 electron fermion
1 up quark fermion
2 down quark fermion
bosons 0 photon boson

fermions bosons
quanta type quanta type
0 electron fermion photon boson
1 up quark fermion NaN NaN
2 down quark fermion NaN NaN
There are no rows in this table
⁠
Inner and outer joins are another way to merge DataFrames. You give a key and axis, and specify the join.
You can also join multiple merged DataFrames.
1
joinedBucket = [fermions,bosons]

joinedBucket = pd.concat(joinedBucket, keys=['type', 'charge'],axis=1, join='inner')

print(joinedBucket)

type charge
quanta type quanta type
0 electron fermion photon boson
1 up quark fermion NaN NaN
2 down quark fermion NaN NaN
There are no rows in this table
⁠

Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.