Analyzing Data Using i2 Analyst's Notebook

Often analysts use IBM i2 Analyst's Notebook for drawing pretty pictures, such as organization charts, but it is capable of so much more! It especially useful for analysts who work with transactional data. Imagine being able to understand money flow, pattern of life, how networks are connected, where points are located, and more, ultimately creating a finished product that management can understand, all from one tool!

These are some of my favorite techniques for analyzing transactional data in i2 Analyst's Notebook. I hope you find them as useful as I do!

Please note this list is not exhaustive by any means. If you have a technique I did not include, I would love to know about it! This post is informal, the data is fictitious, and represents my personal views, not necessarily those of IBM.

Import the Data

First, import structured transactional data (e.g. phone calls or financial transactions) from .xls, .csv or .txt. Be sure to:

• Use multiple multiplicity on the link representing the call between two phones/ accounts. This allows each call/ transaction to be analyzed
• Include both the date and the time in the link properties. This allows dates and times to be analyzed for patterns
• Put the duration of the call or transaction dollar amount on the link label. This allows call or money flow to be analyzed

After the data is imported you can do the following:

ANALYSIS ATTRIBUTES

These provide insight based on properties and metadata associated with entities.

1. Click the Analyze tab and then List Items, then click on the Entities tab
2. Click on the Columns button in the lower left corner
3. Click on the Analysis Attributes tab
4. Select a variety of relevant attributes such as:

• Entity Connections which shows how many unique phone numbers (or account numbers) are linked to the phone
• Entity Links to see how many total transactions that phone (or account) had, regardless if all of the transactions were to one number or many
• Entity Earliest Link to see when the transactions began
• Entity Latest Link to see when the most recent transaction was for that phone in this data set
• Entity Sum Links To to see the amount of time spent on incoming calls (or amount of money deposited to the account)
• Entity Sum Links From to see the amount of time spent on outgoing calls
• Entity Sum Links to see the total amount of time this phone was on calls. 

5. Click OK and these analysis attributes will appear in the rightmost columns in the List Items window. The columns can be moved and sorted.

HISTOGRAMS

Histograms can help identify patterns, peaks and valleys in transactional data.

1. Click the Analyze tab and then the Bar Charts and Histograms button
2. Expand Date & Time
3. Click on “as Histogram” to the right of “Date & Time.” This will show the time range of data in the link chart 
4. Change the scale in the histogram to get different perspectives (e.g. Days, Weeks) 

Consider clicking “as Histogram” next to some of the other options (month of year, day of year, day of month) if the time span of the data is large as these could provide valuable insights 

HEAT MATRIX

The heat matrix provides a contrast to the histogram selected and shows when there is a lot of data, little data, or no data, based on a secondary filter.

1. Click on the “as Histogram” next to “Day of Week” 
2. In the Heat Matrix menu (within the histogram), choose “Date & Time: Hour of Day” to find patterns in the day of the week and the time of the day 
3. When finding time frames of data that are of interest within the histograms, or heat matrix, consider analyzing the subset of data in another link chart. To do this:

• Click the bar(s) in the histogram and/ or cell(s) in the heat matrix of interest which will select the links
• Then hold the shift key and right-click in the chart and choose "Select Connections and Ends"
• Hold the shift key and right-click in the chart area again and choose “Copy with Connections”
• Click the new chart button and paste the selected data. Doing this for smaller data sets will help analyze only the records of interest

SOCIAL NETWORK ANALYSIS

Social Network Analysis (SNA) is a quantitative tool that uses algorithms to study how networks operate, interact and behave. SNA can reveal roles, and structural patterns of entities, as well as the prominence of relationships between them. SNA is used to understand link charts, often large, where it is difficult to determine the location and function of key entities.

To use the SNA centrality measures:

1. Click on the Analyze tab and then the Social Network Analysis button
2. Click the Options tab
3. Check the boxes for Betweenness, Closeness, Degree, and Eigenvector (read the descriptions of each which provide insight into its value)

Go to the Results tab and click the Calculate button

TIMELINE

A timeline can be helpful in providing another view of transactional data as well as visually identifying patterns (such as call chaining).

To convert the link chart to a time line:

1. Right-click in the chart and choose “Select all”
2. Click on the Style tab and choose the Change Representation button
3. Click OK
4. The chart will need to be reformatted a bit – drag the entities from the right to the left side of the chart and go to the Arrange tab and click the Theme Line button

Look for patterns in the data like call chaining

FIND PATH

Find Path can be useful in finding if two entities of interest have any contact, direct or indirect.

1. Select two entities of interest (use the control key)

2. Click the Analyze tab and then “Find Path” button

3. Click OK

4. If a path is been found, click the View tab, and then “Hide Unselected” to clearly see the path

ACTIVITY VIEW

1. Go to the Analyze tab and click the Activity View

2. Right-click on an entity of interest and choose “Add to Activity View”

3. In the Activity View, select the properties to visualize (e.g Date & Time)

4. This will show clusters of activity based on the scale setting.

MAPPING TO GOOGLE EARTH

If the data set contains cell towers with geo-coordinates or other locations with addresses, the plugin to Google Earth can be used to visualize geospatial points.

1. First, select the geospatial data - Go to the Analyze tab and select the Bar Charts and Histograms button

2. Click on the Entity Type filter

3. Click on the geospatial entities (e.g. Address, Cell Tower, Location) and these will be highlighted on the chart

4. Click the Maps button

5. Click “Send x to Google Earth” beneath the Sent Items tab 

CONDITIONAL FORMATTING

Conditional formatting allows you to automatically change the appearance of chart items based on their properties. You can use conditional formatting to emphasize significant information on your chart. It is useful for both analysis and presentation of charts.

Go to the Style tan and select Conditional Formatting. Apply any of the conditional formatting options applicable to the analysis:

a. Color Icons and Links based on the Day of Week

b. Enlarge Icons with the Most Links

c. Hide entities with Only One Link

d. Link Width Based on Value of Transaction

e. Use Country Flags for Telephone with Calling Codes

f. Enlarge & Color Frames by Social Network Analysis Eigenvector

g. Enlarge & Color Frames by Social Network Analysis Betweenness

FIND CONNECTING NETWORK

Find Connecting Network provides identification of a common path between two or more entities.

1. Go to the Analyze tab and click the Find Connecting Network button

2. Select two or more entities of interest in finding a common network/ shortest path

3. Click the Add Selected Entities button

4. Click Find Network button

SELECTION SETS

Selection sets are a mechanism for tracking which items in a chart are associated. 

For example, if a certain ten records are associated with one person, and another five records were associated with a second person, selection sets would make it easy to find these records on the chart, and even see which records are associated with both people.

1. Highlight the items (and entities and/ or links) in a chart to be added to a set
2. Go to the Select tab
3. Click on Selection Set 1 and “Add to Set”
4. Select additional items and either add to additional set, or change a new selection set number and “Add to Set”
5. To find items in multiple sets, use Bar Charts and Histograms, select Selection Set 1, and filter by “Yes” value, then selection additional sets, and filter by “Yes,” then use “Fit Selection to Window” button to find overlap.