Slide 3 Ég
Slide 3 Ég
ENGINEERING
PIPELINES
Part 3: Data Visualization
1
Part 1: Data Formats
Part 3: Data
Visualization
AGENDA – PART 3
• Data Quality
• Interactive Dashboards
• Plotly
DATA QUALITY
4
BECOMING A DATA DETECTIVE
What is wrong with the following data?
5
WHAT IS WRONG WITH THIS DATA?
North America Temperature
6
WHAT IS WRONG WITH THIS DATA?
Candy Bars Eaten in a Day
Frequency
3
Hours Played
0
0 1 2 3
Day 9
WHAT IS WRONG WITH THIS DATA?
Hours Playing Video Games
3
Hours Played
0
0 1 2 3
Day 10
WHAT IS WRONG WITH THIS DATA?
United States Precipitation
11
INTERACTIVE
DASHBOARDS
12
THE INTERNET
The world’s largest system of digital data?
13
A DASHBOARD DATA PIPELINE
A Simple Server
Client
2: Server returns Server
webpage with all
data
3: User’s
computer filters
data for
interactivity
14
DASHBOARDS FOR BIG DATA
A More Complex Server
1: User asks
for information
2: Server
returns
webpage Server Server
Client CPU 4: CPU GPU
3: User requests
requests computation
filtered data from GPU
6: server
5: GPU returns
returns filtered
result
data
15
MAKING A GOOD DASHBOARD
16
PLOTLY DASH
17
PLOTLY DASH
The front end for ML and data science models
https://github.com/plotly/plotly.py
18
MAKING A GOOD DASHBOARD
Div
Date: 1969-07- Show
20 Zeros Div
DatePickerSingl BooleanSwitch
e id=“show-
id=“date- zeros”
picker”
Graph
id=“precipitation-map”
19
MAKING A GOOD DASHBOARD
Div
Div
DatePickerSingl BooleanSwitch
e id=“show-
id=“date- zeros”
picker”
Graph
id=“precipitation-map”
20
MAKING A GOOD DASHBOARD
@app.callback(
Output(‘precipitation-map`, `figure`),
[Input(‘date-picker’, ‘date’), # 1st
input
Input(‘show-zeros’, ‘on’)]) # 2nd
input
def make_graph(first_input,
second_input):
…
21
LET’S GO!
22
28