Anupam Goswami

Cognitive Analytics
 
Cognitive analytics is a type of data analysis that involves the use of advanced technologies, such as artificial intelligence and machine learning, to analyze data and identify patterns, insights, and anomalies that would be difficult or impossible to detect using traditional analytics approaches
Difference between Cognitive and other traditional analytics approaches
Cognitive
Descriptive
Predictive
Prescriptive
uses advanced algorithms and machine learning to analyze data in real-time, with the goal of identifying patterns and insights that are not immediately apparent.
primary purpose of descriptive analytics is to summarize and describe historical data, providing insights into what has happened in the past
primary purpose of predictive analytics is to make predictions about future events or trends based on historical data
give detailed differences between cognitive analytics and predictive analytics
uses more advanced statistical and machine learning techniques to identify patterns and relationships in data.
Uses simple statistical analysis techniques to summarize data, such as measures of central tendency (mean, median, mode) and dispersion (variance, standard deviation).
uses statistical and machine learning techniques to identify patterns and trends in data, with the goal of making predictions about future events.
uses mathematical algorithms and optimization techniques to identify the best course of action, given specific constraints and objectives
can also use structured data, but it is better suited for analyzing unstructured data, such as text, images, and social media posts.
typically relies on structured data, such as data from relational databases or spreadsheets
relies on structured data, such as data from databases or spreadsheets
relies on structured data, such as data from databases or spreadsheets
provides predictive insights, identifying trends and patterns that are likely to occur in the future
provides descriptive insights into what has happened in the past, such as identifying trends and patterns in historical data
generates predictions about future events based on historical data
generates recommendations for action based on the results of predictive analytics
uses machine learning and artificial intelligence to provide automated interpretation and insights, making it faster and more accurate than descriptive analytics.
relies on human interpretation to understand the insights generated from the analysis
relies on human interpretation to understand the predictions generated from the analysis.
generates recommendations for action based on the results of predictive analytics
Used in banking industry to monitor transaction patterns, identify unusual activities, and flag them as potentially fraudulent
used to identify historical patterns of fraudulent activities in order to develop risk mitigation strategies for the future
can be used to analyze stock market data and identify stocks that are likely to increase in value in the future.
can be used to detect fraud in real-time and provide recommendations for specific actions to be taken, such as flagging suspicious transactions for further investigation
In the e-commerce industry, cognitive analytics can be used to analyze customer interactions with a website to identify patterns, such as what products are frequently purchased together or what type of content is more appealing to the customers
can be used to identify trends in customer behaviors and preferences.
can be used to identify customers who are likely to churn, allowing marketing teams to develop targeted retention strategies
prescriptive analytics can be used to identify the most effective marketing channels and messaging based on customer data.
can be used to identify trends in customer behaviors and preferences.
descriptive analytics can be used to analyze the prevalence of certain diseases in a population or identify risk factors associated with certain conditions
Can be used to forecast future health risks or outcomes based on historical data. For example, predictive analytics can be used to identify patients who are at high risk of developing a particular condition, allowing healthcare providers to develop targeted interventions
can be used to identify the most effective treatments for a particular condition based on a patient's medical history and other relevant data.
 

Fleiss' Kappa
Cohen's Kappa
This is a way to measure agreement between 3 or more raters. Used for nominal data (e.g. likert scale).
Therefore this measures agreement between 3 or more dependent categorical samples
Similar to Fleiss’s Kappa This is a way to measure inter rater reliability but for below scenarios:
-          2 raters rate same trial once each or
-          1 rater rates 2 trials (measures agreement of new method with old or over time),
Can be used for any number of raters
Can be used for only 2 raters
Allows for scenario where each rater is rating different items also
Only works for scenario where raters are rating identical items
Assumption includes that raters are chosen independently from larger set
Assumption includes that raters are chosen deliberately and are fixed
Scenarios for use:
5 raters randomly picked from a pool asked to give pass/fail by picking samples randomly from pool (e.g. destructive tests)
Scenarios for use:
2 raters asked to give pass/fail for 20 interview candidates
Have 2 machines for measuring pass/fail of an item’s attribute
 
Condition of random sampling among raters means this is not suitable if all raters are reqd to rate all samples
Conversely not suitable if all samples cant be rated because of cost of test or if its destructive in nature

Violin Plot

This is a way of plotting numerical data which is a combination of box plot and kernel density plot. Like a box plot, this too shows the median (indicated by a white dot), the interquartile range (indicated by the broad black bar running along the plot), the minimum/ maximum (indicated by the thin black line running along the plot) and the outliers.

However, on top of the above summary statistics, the violin plot also shows the data distribution which is especially preferred if the data has multiple modes (reference to the shape of violin). This allows us to see the distribution of the data and especially useful if we want to compare multiple groups.
In the above diagram, the violin plot has 2 wide sections showing that majority data points are grouped around that value. So for example if we want to study the grades obtained by students where there are generally multiple groups or modes (say Grades A and C), the violin plot is better to visualize and compare the data. Another example is if we want to compare heights of people across countries, then again, the violin plot is better. For plot of each country, we would typically observe 2 peaks (for males and females)

Abilene Paradox
 
It states that a group or organization frequently take decisions or actions which is opposite to what most or all of the group wanted to. It indicates a breakdown in how a group communicates - with the individuals incorrectly believing that their own choices or preference are contradictory to those of the group (therefore the fear or appearing as “not a team player”) and doesn’t even share them. This is likely caused by human’s instinctive nature to conform to the group, gap of ownership mindset or perceiving an expert/ heavyweight in the group on the topic or even adequate time to debate. Individuals as a result are afraid of speaking up, resulting in collectively the group remaining silent on the issue.
 
For example, after multiple election defeats members of a political party committee meet to select a new president replacing the existing incumbent. Meeting starts with a coterie of close senior leaders extending supporting for the incumbent to continue while covering up the issue. A senior leader who viewed himself as a contender but didn’t want to come out as too eager, extended his support for status quo. Other individual members who too had discussed privately the need for change but didn’t want to appear to present conflict or lack of public support with the party’s ageing leadership, didn’t raise any objections to the incumbent continuing or even share their actual views about need for change in leadership. Some deferred to the senior party leader earlier expressing support with the incumbent. Even the incumbent who didn’t want to appear to run from responsibility while privately of the opinion for fresh blood, didn’t voice out opposition to his continuing as party president in the meeting.
 
Thus, the group while individually desiring of a change in leadership and backed by data of poor decisions made by the incumbent, collectively decided for status quo. This caused further increase in frustration within the members, with each blaming other members or the party leaders and obviously more electoral defeats.
 
In effect, the issue for which resolution was being attempted got compounded due to the hampered decision making ability, with blame game running rife among the members.
As on any path for resolution, the first step is to identify existence of the problem. Symptoms include:
·         Individuals sharing differing view points or opinions while in private as against when in a group
·         Culture of discouraging members presenting different view points or opinions or taking risks or shooting the messenger
·         Presence of frustration among members towards other groups or leadership.
·         Lack of trust among members of the group resulting in backstabbing.
·         Decision making requiring unanimous committee agreement.
 
Now the step towards avoiding this paradox are:
·         Leadership acknowledge the issue clearly asking for frank opinions without fear
·         Encourage everyone to share their feedback even if different or voice out concerns before decision is taken. Reward the devil’s advocate even if he gets proven wrong.
·         Further before collectively discussing, encourage individuals or smaller teams to brainstorm independently and present options. This would ensure that heavyweights don’t swing the discussion first.
·         Encourage members to listen without passing judgment, enforce culture of brainstorming.
·         Finally avoid being hasty while making key decisions without considering alternatives and studying the impact