Lico Stats¶
This document show the preparation/exploration phase before of the bubble chart representation shown on the main page. The work was done in Python with help of jupyter.
In [1]:
import pandas as pd
import numpy as np
import seaborn
import matplotlib.pyplot as plt
import json
%matplotlib inline
%pylab inline
In [2]:
csv_path = "./va-project/data/lico.csv"
Data Preparation¶
In [3]:
# Read dataset from csv
df = pd.read_csv(csv_path, skipinitialspace=True, delimiter=";")
# Deleting attributes not useful
df = df.drop("diskSpace",1)
df = df.drop("swap",1)
df = df.drop("memory",1)
df = df.drop("distroVersion",1)
df = df.drop("userId",1)
df = df.drop("kernel",1)
df = df.drop("cpu",1)
df = df.drop("country",1)
# Substituting null values and /N with "unknown"
df = df.replace("\\N","unknown")
df = df.fillna("unknown")
df = df[df["numCores"] != "0"]
for a in df:
df = df[df[a] != "unknown"]
Simple data exploration¶
In [4]:
df.sample(10)
Out[4]:
Distribution¶
First 30 distro ordered by number of machines
In [5]:
pylab.rcParams['figure.figsize'] = (15, 6) #Image size
ax = df["distribution"].value_counts().head(30).plot(fontsize=15, kind='bar', color='#7CB518')
ax.set_ylabel("Machines",fontsize=15)
Out[5]:
Class¶
In [6]:
pylab.rcParams['figure.figsize'] = (15, 6) #Image size
ax = df["class"].value_counts().plot(fontsize= 15, kind='bar', color='#F7B32B')
ax.set_ylabel("Machines",fontsize=15)
Out[6]:
Architecture¶
In [7]:
pylab.rcParams['figure.figsize'] = (15, 6) #Image size
ax = df["architecture"].value_counts().plot(fontsize=15,kind='bar', color='#5BC0EB')
ax.set_ylabel("Machines",fontsize=15)
Out[7]:
Cores¶
In [8]:
pylab.rcParams['figure.figsize'] = (15, 6) #Image size
ax = df["numCores"].value_counts().plot(fontsize=15, kind='bar', color='#EF5B5B')
ax.set_ylabel("Machines",fontsize=15)
Out[8]:
CSV to Hierarchical JSON¶
The CSV file is converted to JSON format in order to follow the hierarchy Distribution->Class->Architecture->Cores
In [9]:
def hierarchical_json(df):
order = ["distribution","class","architecture","numCores"]
dist_attr = df["distribution"].unique()
class_attr = df["class"].unique()
arch_attr = df["architecture"].unique()
cores_attr = df["numCores"].unique()
data_dict = {}
data_dict["name"] = "machine"
data_dict["size"] = df.shape[0]
data_dict["children"] = []
data_dict["depth"] = 0
for dist in dist_attr:
dist_dict = {}
dist_dict["name"] = dist
dist_dict["children"] = []
data_dict["children"].append(dist_dict)
dist_df = df[df["distribution"] == dist]
dist_dict["size"] = dist_df.shape[0]
dist_dict["depth"] = 1
class_attr = dist_df["class"].unique()
for class_ in class_attr:
class_dict = {}
class_dict["name"] = class_
class_dict["children"] = []
dist_dict["children"].append(class_dict)
class_df = dist_df[dist_df["class"] == class_]
class_dict["size"] = class_df.shape[0]
class_dict["depth"] = 2
arch_attr = class_df["architecture"].unique()
for arch in arch_attr:
arch_dict = {}
arch_dict["name"] = arch
arch_dict["children"] = []
class_dict["children"].append(arch_dict)
arch_df = class_df[class_df["architecture"] == arch]
arch_dict["size"] = arch_df.shape[0]
arch_dict["depth"] = 3
core_attr = arch_df["numCores"].unique()
for core in core_attr:
core_dict = {}
core_dict["name"] = core
core_df = arch_df[arch_df["numCores"] == core]
core_dict["size"] = core_df.shape[0]
core_dict["depth"] = 4
if(core_df.shape[0] != 0):
arch_dict["children"].append(core_dict)
return data_dict
In [10]:
data_out = hierarchical_json(df)
with open('data.json', 'w') as outfile:
json.dump(data_out, outfile, ensure_ascii=False)
Back to main page.