Descriptive Analytics Explained

• Descriptive analytics interprets historical data to understand business changes.
• It describes using historical data to compare reporting periods within a company or across the industry.
• Common financial metrics include year-over-year pricing changes, month-over-month sales growth, and total revenue per subscriber.
• Descriptive analytics processes raw data to draw understandable conclusions for managers, investors, and stakeholders; it provides a picture of past performance.
• The analytics compare an organization's performance with others in the same industry.
• Performance metrics flag strengths and weaknesses to inform management strategies.
• Context is needed to understand reports; a $1 million sales report requires knowing if that is a decline or increase.
• Larger context, including targeted growth, is required for informed views of sales performance.
• Descriptive analytics is a core part of business intelligence and has industry-specific and broadly accepted measures.
• Companies use these analytics to gain valuable insights into their performance and position in the market, and to determine financial trends.
• It helps companies understand their operational efficiency and identify areas for improvement, including motivating different teams to implement changes.
• Data aggregation and data mining are two primary methods of data collection used for descriptive analytics.

How to Implement Descriptive Analytics

• Identify metrics to analyze and the time frame, then find all the data from internal and external sources, including databases.
• The identified data is compiled and formatted and datasets and figures are analyzed with different tools.
• All data is presented to stakeholders with visual aids like charts and videos for insight into the company's direction.
• Return on invested capital is a form of descriptive analytics, created by taking net income, dividends, and total capital, then turning it into an understandable percentage.

Advantages

• Descriptive analytics disseminates information and helps stakeholders understand complex ideas through visuals.
• Stakeholders see how a company compares to others by production costs, revenue streams, and product offerings.
• Companies see areas for improvement in their own business plans/models.

Disadvantages

• These analytics do not help understand what to expect in the future, to account for market forces, or to assess variables that may affect them in the future.
• Stakeholders may choose favorable metrics, creating bias that affects the perception of profitability, ignoring areas that require change.

Competing Analytics Methods

• Newer fields emphasize predictive, prescriptive, and diagnostic analytics to model outcomes and suggest actions that maximize positive outcomes while minimizing negatives.
• Predictive attempts to make predictions through statistics and modeling, using current and past data to determine the likelihood of similar future outcomes.
• Employing predictive can help companies identify and address inefficiencies, and find better ways to utilize resources.
• Prescriptive analytics allows companies to analyze data and find what needs to be done to achieve specific results, while also considering past and current performance.
• Stakeholders using prescriptive can better make decisions across any timeline and determine investment in R&D, product offerings, or whether to enter a new market.
• Diagnostic analytics determines why a trend exists, manually or with computer software, and figures out the root cause of events to make changes.

Applications of Descriptive Analytics

• Companies can draw comparisons with other reporting periods to identify inefficiencies in their operations and make changes for the future.
• It answers the "What happened?" question, while Predictive answers "What will happen?" by using historical data to figure out how to improve.
• Predictive helps companies understand how changes will impact future performance, and these types work together.
• Companies measure audience engagement through social media or analyze financial metrics.
• Measuring social media engagement reveals data on which campaigns or product launches lead to traffic to their sites and referrals.

Algorithms

• Clustering groups similar data points and identifies patterns.
• Association rule mining unveils links between variables and items, which is useful in market basket analysis and recommendation systems.
• Time series gauges patterns, trends, and seasonality in time-dependent data.
• Text Mining & NLP analyses sentiment, topics from unstructured text data (customer reviews, social media).
• Decision trees make hierarchical structures representing decision rules to classify data and highlight key features.
• Geographic information systems are used to analyze spatial data by mapping patterns and trends to specific locations.
• Regression models relationships between dependent and independent variables.
• Data mining identifies unusual patterns.
• The use of specific algorithms is determined by the type of data, the analysis objectives, and the industry and application context.

Descriptive Analytics Process Steps

• Data is collected from sources like databases and spreadsheets.
• Identifying and resolving missing values, inconsistencies, duplicates, and outliers.
• Analysts explore the data using summary statistics, data visualization, and data analysis to identify patterns.
• Segmentation divides datasets into subsets based on demographics, geography, and time for focused analysis.
• Summary measures show averages, percentages. Key performance indicators evaluate business process, product or service.
• Historical trends are analyzed to understand how variables/metrics have changed over time for patterns and seasonality.
• Findings are communicated through reports or dashboards with summary statistics, visuals, and descriptions.
• Descriptive analytics requires continuous data monitoring and updates to capture patterns and trends; monitors sales data, and updates analyses.

Examples

• Analyzing prior sales data can identify top-selling products, impacts of pricing strategies, sales channels, and regions.
• By assessing sales data, one can look at a drop in category sales, and investigate causes (changing preferences or increased competition), or by examining sales in the online sector.
• Segmentation improves marketing personalization and retention.
• Descriptive analytics can segment customers by demographics, purchase behavior, and engagement to tailor marketing strategies.
• Data leads to insights, optimized website operation, and higher conversion rates and user experience.
• Examination of manufacturing production data can lead to optimizing resources, staff, or processes.
• Financial insights are gained into revenue, expenses, and profitability.

Data Collection: Terms and Methods

• Data is a tool that helps people reach a sound conclusion; it understands socio-economic problems.
• Data collection gathers information from multiple sources to find answers to research problems, evaluate outcomes, etc
• Analysts must ask self what the purpose is; what kind of data, plus what methods and procedures to collect with.
• There is qualitative (descriptions) and quantitative data (numbers).
• Data helps investigators understand a problem by providing required information as primary or secondary data.
• Investigators conduct the statistical inquiry and enumerators gather and provide statistical information for data collection.
• Respondents are people from whom data information is then collected.
• Surveys collect information from individuals to describe usefulness, quality, price, and kindness asking about a product/service.

Methods of Collecting Primary Data

• Direct personal investigation obtains data personally from the source.
• Indirect oral investigation collects data orally from someone other than/indirectly connected to the person with the information.
• Information from local sources appoints correspondents who collect data across various areas.
• Information is collected via questionnaires, mailing, and enumerator's methods.

Primary vs. Secondary Data

• Primary data is collected directly from first-hand sources for a specific research purpose though methods including surveys, interviews, experiments, observations, and focus groups.
• Provides current and specific information with accuracy and control; Observer bias can influence the results; The investigator may influence subjects' behavior.
• Customer satisfaction surveys and market research suits online, paper, or face-to-face questionnaires that collect data with a questionnaire.
• Observations record behaviors/events.
• Studying "product user" interactions is good for assessing classroom dynamics, plus wildlife behavior monitoring.
• Experiments analyze what happened, and the cause and effect can be artificial, and limit the ability to generalize the findings.
• The investigator analyzes the efficiency of drugs or marketing campaigns. Gathers with a "moderator" to gather feedback, thought, emotion, and/or insight.

Secondary Data, Sources, and Selection

• Secondary data is collected, processed, and published and saves time/resources.
• Can be collected through different published and unpublished sources such as government publications of statistical data, or published data related to health and education plus, newspapers and magazines providing statistical data.
• Research institutions publish activities/findings, but secondary data requires adjusting to suit the objective and lacks the origin quality and is lower in cost to collect.
• Unpublished sources with data is in the form of research work or records maintained.

Data-Driven Decision Making

• DDDM uses data and analysis to make business decisions, using customer feedback and trends, collect processes enabling businesses to find success.
• Generates real-time insights to optimize performance and test new strategies and provides a solid foundation, which reduces uncertainty.
• Results come from customer engagement and satisfaction, and better strategic planning, using extensive data.
• It is used to personalize experiences, product suggestions or pricing strategies and reduce customer churn, the platform is driven using algorithms.
• Financial institutions use machine learning algorithms to identify/prevent fraud, while utility companies use them to predict energy consumption.
• Data insights formulate realistic/strategic plans.
• E-commerce retailers identify untapped customer segments and develop services to identify markets.
• By analyzing sales data, organizations discover specific products that had a spike before an event.

Decision Making

• Data-driven decisions minimize bias and increase objectivity.
• Implement debiasing techniques, and raise awareness of biases to increase transparency.
• Implement and measure the impact once you: define objectives, identify and collect data, explore data by cleaning its data and see results, and analyze with methodologies or patterns.
• Key findings are reviewed in the context to form actionable insights and will drive business success.
• Implement and evaluate resources allocation.

Challenges of Data-Driven Decision Making

• Organizations avoid data with quality.
• Data illiteracy can lead to misinterpretations and sub-optimal decisions, so provide training while over reliance on historical data can be problematic,
• Confirmation bias arises as decision-makers may interpret data selectively.
• Data types support preconcieved notions while Neglecting data security poses a risk.
• Descriptive analysis summarizes the history of past performance (sales).
• Diagnostic analysis determines why events occurred, mining and identifying correlations to uncover the root causes (drop in sales, customer complaints,).
• Predictive analysis is used to predict sales and customer relation management.
• Prescriptive will suggest the best course of action, from which is derived supply chain optimization.
• Exploratory analysis helps identify markets, and cluster and dimensions data.

Data Preparation and Cleaning

• Data cleaning, or data scrubbing, identifies and corrects errors and inconsistencies in raw data.
• Processes address duplicates, missing values, syntax errors, and structural errors, as well as securing it.
• Organizations with clean data make reliable decisions and respond to changes.
• Cleaning is essential for data science and converts format for analysis.
• Underpins the success of AI and ensures machine learning algorithms, leading to robust predictions.

Benefits of Cleaning Data Analysis

• Informed decision-making aligns good quality data with business goals.
• Improves productivity, cost efficiency, data compliance, enhanced model performance, and data consistency.
• Data assessment reviews a data set to identify quality issues to standardization.
• A common discrepancy is the date format (MM-DD-YYYY vs DD-MM-YYYY).
• Outliers' extreme values distort analysis.
• Data deduplication is a streamlining process that reduces redundant data when the data entry is repeated twice.
• Missing values are values not present and professionals might replace missing data, otherwise known as data imputation.
• The final review verifies that the data is clean and often uses manual inspection.

Guiding Principles

• "Garbage in, garbage out" is when Data Analysts will get unreliable results.
• Data cleaning enhances efficiency and reduction while well-cleaned data streams processing.
• Feature engineering transforms into a more suitable format.
• Scaling is crucial when different features have vastly different sales and may include min-max scaling to transform based on normal distributions.
• Encoding categorical data is done through one-hot that converts 'red', 'blue', 'green', can be converted into three features, that are all colors.
• Advanced cleaning includes automation and ML data.
• Machine learning can be leveraged to refine data that automates cleaning and predicts modeling.
• Data is prepared through altering the numerical columns in the dataset to a standard scale through data normalization.
• Normalization is used to arrange the data and is a scaling method to reduce duplication.
• Used to remove characteristics, normalization and standardization occur.

Data Aggregation Steps

• Data aggregation streamlines performance and provides a high-level view; data is collected in a central repository.
• Requires data loading, processing and summarization.
• Data summarization is a process that Reduces large data sets, making them more concise, while retaining essential information.
• Benefits come in the form of improved analytics, and decision-making, time savings, and increased accuracy.
• Common techniques include aggregation and data sampling, by Selecting a subset of the points from a data set.
• Clustering points to determine what the right summarization technique to make decision in time.
• Use filtering techniques for data summarization to create a concise summary, the technique is applied to focus only on a category or region.
• Sampling, data Visualization, dimensionality reduction and text Summarization.

Additional Methods of Summarizing

• Mean, median, and more allow a data summarization.
• Range is measured in the difference between the highest and lowest values.
• Visuals are key - bar chart, histogram and box plots.
• Use frequency tables for data Summarization by organizing categorical data, the table then displays number of occurrences.
• When creating keep categories accounted for.
• Tables and pie charts simplify categorical.
• Moving averages and trend analysis smoothens the time.
• Data analysis may give Business insights across scientific research.

Data Visualization

• Utilizes graphics (charts, plots, infographics, animations) for easy complex data understanding.
• May convey organizational structure. Used to generate ideas, illustrate and explain processes, aid in discovery, and serve everyday needs.
• Early use was in navigation while dashboards are used to report performance metrics.
• Visuals include: Tables, Pie and bar charts, Line charts, and Scatter plots.
• Heat maps as a graphical representation that helps with behavioral data.
• Provide general background, identify the audience, choose an effective visual; keep it simple.

Exploratory Data Analysis or EDA

• Used by data scientists and helps best manipulate data.
• EDA can reveal data beyond formal testing, that may help the validity of the statistical techniques or create new hypotheses.
• One can Generate Hypotheses, Validate Assumptions or Identifying Data.
• Important factors include: Center, Dispersion, Distribution, and Visualisation
• One may then visualize and determine the types of EDA data by identifying whether it is univariate, or multivariate graphical or non-graphical

Exploratory Data Analysis Language

• The most common programming languages are Python and the "R" language.
• The data is gathered to a specific criteria or through functions for analysis.

Data Analysis Software Applications

• Excel is a common tool for data analysis and analysis, such as calculating with the "analysis tool pack."
• It includes graphing and functions for data performance.
• Python, routinely ranked as the most programming in the world, use it to streamline models, visualize, and analyze data using built-in data analytics tools.
• A key appeal to professionals is the amount of libraries such as Panda.
• tableau is primarily for business analytics and intelligence due to the easy seamless turning of the data.
• MySQL, used for websites is open and secures what's happening .
• sas is used to retrieve an intuitive graphical interface(GUI), that enables and creates