BUSINESS INTELLIGENCE, BIG DATA AND MACHINE LEARNING INTRO

docx

School

California Polytechnic State University, San Luis Obispo *

*We aren’t endorsed by this school

Course

420

Subject

Information Systems

Date

Oct 30, 2023

Type

docx

Pages

Uploaded by CaptainWaterStork35

BUSINESS INTELLIGENCE, BIG DATA AND MACHINE LEARNING INTRO Computing History  Moore’s Law: computing power will double every 2 years.  Memory and storage cost has decreased.  Massive increase in ability allowed storage of data, processing, and inference to become much more accessible for all businesses and has driven the rise of business intelligence. Data Driven Business Strategy: Business Intelligence R Cheaper and faster: data storage, processors, memory R There is prolific data collection , because of the ubiquity of the network connectivity R Data Management has become far more sophisticated R Rise of cloud computing , which enables all of the collection and processing to be done much more efficiently and effectively. R Together all of this has contributed to the rise of AI and ML. DATA DRIVEN DECISION MAKING Business Analytics describes Data Driven Decision Making Big Data R Volume: Includes records, transactions, files etc. R Variety: Data can be structured, semi structured, unstructured. R Veracity: Most critical component. The trustworthiness, authenticity of analysis, accountability, and availability for use of the data. Central for quality decision making (& AI). R Velocity: how fast data is arriving; in batches, in real time or streams. R Some include Value as the 5th component to define Big Data. Knowledge comes from ML, AI, Data Mining, Deep Learning Artificial Intelligence (AI) encompasses the whole bit of knowledge. Machine Learning (ML): Statistical and algorithmic methods for data analysis (regression, decision tress, clustering, algorithms) o Supervised Learning: along with the data an outcome is provided by the human. Train algorithm to Finds patterns associated with certain outcomes. o Unsupervised Learning/Data Mining: the algorithm finds patterns on its own

o Deep Learning: ML algorithm that emphasizes hierarchical learning* using neural networks. Can be supervised, unsupervised or semi supervised. Problem Types:  Classification: binary, multiclass, multilabel  Numeric Prediction: e.g. profit to expect this quarter ML Methods Learning Types  Supervised  Unsupervised  Semi supervised: algorithms undertake certain pattern recognitions; humans validate some of these findings. Machine learns from data and human input. CLUSTERING Clustering: most versatile classified algorithms which relies on identifying clusters of data points and what allows defining of this cluster.  Most common used ML algorithm,  Commonly used as a way to reduce data.  Identify clusters through variables and after identifying clusters, labels are assigned.  k-Means Cluster: Given a data set, the goal is to identify k-sets of means. Evenly spaced points keep converging till it starts getting to a center of a set of data points. Reduces until centers of clusters are identified. Limitations: when it’s not able to divide the data up in a proper cluster.  Density-Based Spatial Clustering of Applications with Noise (DBSCAN); Finds clusters in any given data set by continuously scanning on adjacent density of the class of data points that you are trying to group together. Clustering Types o Hierarchical: clusters belonging to a higher-level cluster o Overlapping: i.e. customers who like romance movies and wine. o Subset: similar to hierarchical, but typically tends to be 2 levels or less.

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version