The methodological and iterative assessment of the information given by an organization is business analytics that emphasizes the statistical assessment. It is used primarily by companies that are interested in verdicts based on results. It deals primarily with market intelligence and statistical analysis. It is used mainly to gain insights into business decisions and to improve business processes. This paper addresses some of the terminology used in corporate analysis, including logistic regression, case-based reasoning and k-means clustering. Logistic Regression
In business analytics, logistic regression is the fitting regression evaluation to conduct when the dependent variables are binary (dichotomous). It is a predictive evaluation and is used to define data as well as explain the connection that exists between a dependent dichotomous variable and ordinal, interval, ratio-level, or nominal independent variables (Statistics Solutions, 2017). Some of the key assumptions of logistic regression include:
The result has to be discrete or rather the dependent variables ought to be binary in nature
No outliers should be in the data
The predictors should not have a high intercorrelation
The dependent variables are stochastic events
The resultant variable is located in the first box as the other predictors are placed in the covariates box.
Comparison of Artificial Neutral Network and Logistic Regression
An artificial neural network (ANN) is a computing structure comprising of several simple, highly connected processing rudiments that process data by their dynamic condition state response to the exterior inputs. Logistic regression and Artificial Neutral Network are algorithms associated with the organization of problems where discrete values are probable. While the ANNs have a specific structure having a single input, hidden layer, and an output layer, the Logistic regression has distinct variable classification with several individuals using the softmax and sigmoid function to multiclass classification difficulties. Both processes have the same issue i.e. establishing the most accurate value for their factors. In most cases, gradient descent is utilized in logistic regression to reduce the cost of each function while also bettering the values. On the other hand, back propagation is used in Neutral Networks to identify the parameters (Hon-Yi Shi, 2012).
In this process, old experiences are utilized in the comprehension and solving of current problems. An individual can remember a previous circumstance that is similar to the new issue and use the knowledge to address the problem. A good example can be drawn from the fixing of a vehicle or other types of machinery; whereby an individual remembers how the previous issue was solved and applies it to the current one. It may also mean using old findings to critique new measures (Ashok K. Goel, 2017).
Advantages of Decision Trees
Decision trees are elements used by managers to analyze future choices. They create a visual representation of the likely results, follow-up decisions, and rewards. One of the key advantages of the decision tree is the brainstorming results. They assist an individual in thinking about the probable results as well as the consequences; hence, enhancing the accuracy of the decisions. It adds transparency to the process as the graphical portrayals make it easier to compare the different alternatives (Nayab, 2011).
The decision tree is quite versatile since it allows for customization. As such, it makes it easier for the users to diagnose problems and rectify the same. For instance, a technician can easily identify a mechanical failure and troubleshoot the repairs. In the business perspective, managers can use the same to evaluate the corporation’s decision-making process. The decision making a tree is also flexible handling items with different categorical features, unlike other tools that need comprehensive quantitative information (Nayab, 2011).
K-means is a learning algorithm utilized in solving clustering problems. The working procedure is as follows: a simple manner of representing data via a specific cluster number i.e. k clusters is provided. K centers for all the clusters are defined and placed in a cunning form since different locations lead to different outcomes; hence, they are placed far apart from each other. From there, each point in a particular data set is linked to the closest center. The first step is deemed to be complete once all the points are linked to the respective centers. The next phase entails the recalculation of K centroids and the establishment of a new binding emanating from the same data points as well as the closest new center. This aspect forms a loop that makes the K centers to alter their location until the centers stop moving (Trevino, 2016).
The terms elucidated above are primarily used in business analytics to comprehend the different concepts entails in the same. Logistic regression, artificial neutral network, case-based reasoning, decision trees, and k-means clustering are vital concepts in the comprehension of business analytics.
Ashok K. Goel, &. B.-A. (2017). What’s Hot in Case-Based Reasoning. 1-3.
Hon-Yi Shi, King-Teh Lee, Hao-Hsien Lee, Wen-Hsien Ho, Ding-Ping Sun, Jhi-Joung Wang, & Chong-Chi Chiu. (2012). Comparison of Artificial Neural Network and Logistic Regression Models for Predicting In-Hospital Mortality after Primary Liver Cancer Surgery. Retrieved from http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0035781
Nayab, N. (2011, July 2). Decision Tree Forests. Retrieved from http://www.brighthubpm.com/project-planning/106000-advantages-of-decision-tree-analysis/
Statistics Solutions. (2017). What is Logistic Regression? Retrieved from http://www.statisticssolutions.com/what-is-logistic-regression/
Trevino, A. (2016, June 12). Introduction to K-means Clustering. Retrieved from https://www.datascience.com/blog/introduction-to-k-means-clustering-algorithm-learn-data-science-tutorials