Automating our work away One consulting firms experience with - - PowerPoint PPT Presentation

▶

Apr 09, 2023 276 likes •549 views

Automating our work away One consulting firms experience with RMarkdown Finbarr Timbers 1 2 Outline 1. About Darkhorse Analytics 2. Problems we faced 3. How we solved those problems 4. How you can improve R(Markdown) adoption 3 4 5 6

SLIDE 1

Automating our work away

One consulting firm’s experience with RMarkdown

Finbarr Timbers

1

SLIDE 2

2

SLIDE 3

Outline

1. About Darkhorse Analytics
2. Problems we faced
3. How we solved those problems
4. How you can improve R(Markdown) adoption

3

SLIDE 4

4

SLIDE 5

5

SLIDE 6

6

SLIDE 7

Problems

7

SLIDE 8

Problems

1. It takes a lot of time to replicate, which is a problem when

rework happens (>50 slides!). We’ve had 3 month projects with 1 month consumed by rework.

8

SLIDE 9

Problems

1. It takes a lot of time to replicate, which is a problem when

rework happens (>50 slides!). We’ve had 3 month projects with 1 month consumed by rework.

2. All styled differently. We need to edit every plot individually.

9

SLIDE 10

Problems

1. It takes a lot of time to replicate, which is a problem when

rework happens (>50 slides!). We’ve had 3 month projects with 1 month consumed by rework.

2. All styled differently. We need to edit every plot individually.
3. No documentation on how to replicate.

10

SLIDE 11

Instead? R!

11

SLIDE 12

What we did:

12

SLIDE 13

What we did:

1. Use RMarkdown to write template (“best practice”) slides.

13

SLIDE 14

What we did:

1. Use RMarkdown to write template (“best practice”) slides.
2. Made the slides accessible to the company (we used Git +

Bitbucket).

14

SLIDE 15

What we did:

1. Use RMarkdown to write template (“best practice”) slides.
2. Made the slides accessible to the company (we used Git +

Bitbucket).

3. Made it easy to run in other environments (we used installation

scripts; moving to Docker)

15

SLIDE 16

What we did:

1. Use RMarkdown to write template (“best practice”) slides.
2. Made the slides accessible to the company (we used Git +

Bitbucket).

3. Made it easy to run in other environments (we used installation

scripts; moving to Docker)

4. Iterate (we used Git).

16

SLIDE 17

What we did:

1. Use RMarkdown to write template (“best practice”) slides.
2. Made the slides accessible to the company (we used Git +

Bitbucket)

3. Made it easy to run in other environments (we used installation

scripts; moving to Docker)

4. Iterate (we used Git).
5. Publish as a package (yet to happen).

17

SLIDE 18

R:

darkhorse_histogram = function(df, x_var, x_breaks, x_label, y_var, y_breaks, y_label, title, col="#E69F00") { ggplot(df, aes_string(x = x_var, y = y_var)) + ggtitle(title) + geom_bar(fill=col, stat = "identity") + geom_hline(yintercept= 0, colour=col) + darkhorse_theme() + ylab(y_label) + xlab(x_label) + scale_y_continuous(breaks=y_breaks) geom_hline( yintercept=y_breaks[2:length(y_breaks)], color="white") + scale_x_continuous(breaks=x_breaks) }

18

SLIDE 19

Go from. . .

19

SLIDE 20

To. . .

2 4 6 8 10 12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Minutes from Home to Station Headcount

Part Time Responders

20

SLIDE 21

Also: SQL

SELECT * FROM (SELECT CASE WHEN Call_Type NOT IN ('Medical', 'Fire', 'Fire Alarm Ringing ELSE Call_Type END AS CallType, CallID, Year FROM ERA_TFS.dbo.Calls_Template WHERE Qualified_Event = 1) t PIVOT (COUNT(CallID) FOR Year IN ([2012], [2013], [2014])) pvt

21

SLIDE 22

Vehicle Incident Fire Other Fire Alarm Ringing Medical

20000 40000 60000 2012 2013 2014

Calls

22

SLIDE 23

Lessons

1. Make it great.

23

SLIDE 24

Lessons

1. Make it great.
2. Make it easy.

24

SLIDE 25

Lessons

1. Make it great.
2. Make it easy.
3. Make it profitable.