In: Nursing
What are some pros and cons to data mining? Provide an example of when data mining was used and the outcome provided an incorrect assumption or issue. How can these types of situations be avoided in the future?
answer: Data mining is a process by which the available data is sorted out to make the result analytical and obtain Interpretation for future change.
data available(unsorted data)-----analyze(data that makes sense-sorted data)-----interpret------implement changes.
no | pros to data mining | cons to data mining |
1. | It helps to study the past trends and provide interpretation. |
Future does not always follow the past. There is always change in the trend and interpretation based on the past data not always be correct |
2. |
It is often analysis of the objective data available by using data analysis tools. |
it often does not take into consideration the subjective data that is the basis or foundation behind every decision or behaviour that provides objective data and should include subjective data for results to be correct. |
3. |
Data mining focuses on the analysis from the data(representative data) that is available for collected data at any point of time from the available records. Its based on the data available to study future trends based on already available data and helps industry marketing,research,policy formation and profitability |
The available records may not be the only data that is relevant to analyse a trend and just a part of the entire data, that is correct sample data to make the prediction may be bigger. New policies are formed based on the analysis of the representative data.Since the representative data is only a subset of the whole data, the analysis and predictions are partly biased due to collection data bias. Data mining focuses on data that is available and not the data that is needed and hence is not always accurate. |
4. | It has the advantage of promoting profitability through perfect predictions by observing the current trends and analyzing them | it is expensive to analyse the data and also there are security and privacy issues concerning data which are difficult to handle. |
5. | Its tries to correlate the data to an mathematical observation and details and tries to predict the future behaviour so as to drive change for betterment. | it does not take into account that observed mathematical is borne out of complex human behaviour and nature ,is multifactorial and hence the prediction could be restrictive in nature and could be flawed. |
6. | it is correlational interpretation of data | it is not causative interpretation of data |
An example of when data mining was used and the outcome provided an incorrect assumption or issue
Woolworths a grocery chain in Australia that had a rewards Loyalty program in tie up with another company Qantas in order to understand the effect the Loyalty program it had,it employed a data analysis team which used data mining tools to predict that cash back instead of card loyalty will help in in promoting or boosting the sales.The net result of implementation of the data mining result was a decrease in the sales and the customer satisfaction. Data mining prediction proved to be wrong and resulted in declining sales for Woolworths. This is an example to show that how the results of data mining can go wrong.
.Google flu employed a search engine for studying the trends of influenza disease in the the outbreak period by searching the IP addresses of the people who searched flu-like symptoms on the internet. Since the flu symptoms are the subset of symptoms that could occur in a variety of diseases, its accuracy in predicting the trend in influenza was flawed. In the influenza outbreak of 2011 to 2013, it was unable to accurately predict the rates of existence of Influenza and differentiate it from influenza like diseases which have an overlapping cluster of symptoms with influenza using data mining.
The ways in which these types of situations be avoided in the future:
Collection and analysis of Subjective data along with objective data by obtaining a real-time feedback from the customers or the client directly connected with the service that is being analyzed for data using data mining will help to give a global interpretation of the results.This can be done in the form of questionnaires and surveys and should be considered crucial to data mining.
Obtaining information from the non-users of the services will help to remove the bias that is caused by data mining from only the available data or the sample users data,to the the data mining that targets the entire population
Implementing the changes predicted by data analysis for a subset of population and observing the feedback rather than radically changing the the information obtained from data mining for the entire business process or practices can help to to study the effect of change and modify accordingly (A/B testing)This provides validity to the change that is suggested by the data mining result