Upload
anna-smith
View
34
Download
2
Tags:
Embed Size (px)
Citation preview
1
But can your data tool make your data team
work smarter?
2
A typical member of a data delivery teamis a single point of failure.
Why?
3
Because she has unique knowledge of• Specific data domains• Specific processes that pipe the data to
a specific client• Specific applications for specific clients
to use the data
4
She has this unique knowledge because she’s the one who personally created the experience based on her• Personal style of cleaning data, and only to a
minimum as driven by the specific client• Personal naming conventions, development style• Personal coding ability • Personal level of need for documentation
• —of course all team members will be different
5
Your team should be using data that is standardized—
There are at least 4 major data suppliers to your data delivery team:• The upstream, even bigger data team• The team who developed that custom app for your line of business• The other team who developed that other custom app for your line of business• That major core corporate system that is now in the process of being replaced
• But with all these suppliers, can you honestly hope that data will
arrive the same way always?
6
On your current corporate-approved platform:• There is a data handling limit (2 gigabyte), compared
to the actual size of the data• Or there is no data handling limit, but usability is so
clunky that users prefer to code over drag-and-drop (and they are proud of it)
• There are several versions of the corporate-approved platform in production
• Or if you have more than one platform, the dueling platforms have a you-say-potAYtoe-I-say-potAHtoe problem.
7
Natural Personal Variation
Standard Corporat
e Platform Constrain
ts
Natural Data
Variation
These 3 elements combine, creating the existing work environment
for your data delivery team
8
Natural Personal Variation
Standard Corporat
e Platform Constrain
ts
Natural Data
Variation
There other are side effects emanating from this environment
single points of failure
instability
furniture
inefficiency
behind the times in terms of techniques
A data delivery process that the client does not use to make decisions
9
Natural Personal Variation
Standard Corporat
e Platform Constrain
ts
Natural Data
Variation
Some things you can’t change. What can you change to eliminate the side effects?
single points of failure
instability
furniture
inefficiency
behind the times in terms of techniquesthe tool
10
You’ve always said the corporate-approved platform was “free”, but how expensive are the
negative side effects?
11
What positive side effects do you want?
What features would yield these?
12
1Eliminate
Data Chunking 2
Less Data
Processing or
Coding
3 More Work
Recycling
4Data
Consistency a Done
Deal
5 Intra-Team
Communi-cation
6 Unrestriced Data Shaping
7 Higher Order
Functions
13
1 Eliminate Data Chunking
A better tool would allow you to handle any size data.• Then you wouldn’t have to build and manage
processes to chunk it• Then you wouldn’t have to manage data on either
side of the chunking line• You would not need to make tables and take up so
much server space
14
Here is the property sheet for Input Data
Here is the Input Data icon
Notice all the supported file types
15
2 Less Data Processing
If you did not have to habitually make tables:• You wouldn’t employ other action queries like
append, delete, and update• Then you wouldn’t have to wait for all that
processing• The standard tool wasn’t designed or optimized
for that kind of processing anyway
16
Here is the Browse icon
Here is the view of the data.
When you are finished developed, remove or disable Browses for optimization
17
2 Less Coding
Coding is necessary the further up the data stream, because it is physically necessary.
But coding is a hindrance the further down the data stream, as more and more human factors come in to play.
18
3 More Work Recycling
A better tool would package macros so that they can be exported and imported from project to project, regardless of domain• Team members swapping packages• Packages available from the internet• All new work builds on previous work, lowering
the risk of furniture
19
You can do Events.
Here is the property sheet for an entire module.
Here is where you import or export an entire module
You can download someone else’s module off the internet. You are not transporting data, just a program.
20
4 Data Consistency a Done
DealInstead of enforcing data consistency across 4 or more data offices, you need a tool with which you write the
once-and-for-all program to enforce data consistency within your walls, regardless of which data source or which team member.
21
Also, you can gauge efficiency and reasonable output of your module.
This is the property sheet for Dynamic Input
Here is the Dynamic Input icon
22
5 Intra-Team
CommunicationA better tool would use standard graphics while applications were created in such a way that documentation was inherent.• Drag and drop workflows• Annotation, even default annotation• Icons, not uniquely named queries, sets, or
programs
23
This is a Union icon
This Filter icon, with possible T or F outputs, could have been inserted as an afterthought.
This note on this Sort icon was automatic, too.
This is a Join icon with possible Left, Inner, or Right join options. Make the flow emate from the correct one. Check your decision with a Browse.
24
6Unrestricted Data
ReshapingA better tool would go further than field manipulation.• Not just crosstab, but also transpose• Apply a formula to multiple fields• Do operations on previous row or next row
25
Transpose
Cross Tab
The Multi Row Formula function means you can manipulate based on previous or subsequent rows.
26
7Higher Order Functions
A better tool would offer higher order functions for everyday data situations, saving time.• Cleaning? Try a Parsing function, then a Regular
Expression (pattern matching function, but then a string function only as last resort
• Matching? Try Fuzzy Matching were “Rob Snow” and “Robert Snow” match
27
This is the language for Regular Expressions. It is vast, and there are many “RegEx” icons for different little jobs.
Parsing text to columns is an everyday data chore which can be done with string functions, but not nearly as efficiently as a function built specifically for the job.
1Eliminate
Data Chunking 2
Less Data
Processing or
Coding
3 More Work
Recycling
4Data
Consistency a Done
Deal
5 Intra-Team
Communication
6 Unrestricted Data Shaping
7 Higher Order
Functions
*or love, as needed or preferred
Team actually a team
Data stable and big
glory*
Efficiency
Advanced analytics
Low risk of furniture
29
1Eliminate
Data Chunking 2
Less Data
Processing or
Coding
3 More Work
Recycling
4Data
Consistency a Done
Deal
5 Intra-Team
Communication
6 Unrestricted Data Shaping
7 Higher Order
Functions
30
Thank youAlteryx ToolsAlteryx
Functions