Fiche de révision : Fundamentals of Data Analytics

📋 Course Outline

1. Types of Analytics
2. Data Collection and Storage
3. Descriptive Statistics
4. Predictive Analytics
5. Prescriptive Analytics
6. Measures of Central Tendency
7. Measures of Spread
8. Sampling Methods
9. Types of Data
10. Data Measurement Scales
11. Dispersion Measures
12. Outliers and Influence

📖 1. Types of Analytics

🔑 Key Concepts & Definitions

• Descriptive analytics: The process of analyzing data that has already occurred to summarize and understand past events, providing insights into what has happened. It involves techniques like calculating averages, medians, modes, and measures of spread (variance, standard deviation).
• Predictive analytics: The use of mathematical and statistical models to analyze past data in order to predict or classify future events. It involves building models that can forecast outcomes based on historical patterns, applicable to both numerical and categorical predictions.
• Prescriptive analytics: The process of using data assets as a guide to influence or change future outcomes. It involves creating data-driven solutions to control or optimize events, guiding decision-making to achieve desired results.

📝 Essential Points

• Descriptive analytics provides a historical view, focusing on summarizing data through measures like mean, median, mode, and measures of dispersion (variance, standard deviation). It is also called inferential statistics.
• Predictive analytics relies on building models that use past data to forecast future events, enabling classification and prediction of outcomes. It is essential for anticipating trends and behaviors.
• Prescriptive analytics goes beyond prediction by actively suggesting actions based on data insights to influence future events, aiming to optimize results.
• These three types of analytics form a continuum: descriptive explains what happened, predictive forecasts what might happen, and prescriptive guides what should be done to achieve specific outcomes.

💡 Key Takeaway

Understanding the distinct roles of descriptive, predictive, and prescriptive analytics helps in selecting appropriate data analysis techniques to interpret past data, forecast future events, and actively influence outcomes.

📖 2. Data Collection and Storage

🔑 Key Concepts & Definitions

Data analytics process: The systematic approach of collecting, organizing, and storing data to answer specific questions, enabling informed decision-making and deriving insights (source content).

Data-driven recommendations: Business suggestions or strategies formulated based on analysis of collected data, where business data directly answers business questions (source content).

Population parameters: Numerical characteristics that describe an entire population, such as the population mean or variance, representing the true values of the whole group (source content).

Sample statistics: Numerical summaries calculated from a subset (sample) of the population, used to estimate population parameters, such as sample mean or sample variance (source content).

📝 Essential Points

• The data analytics process involves collecting relevant data, organizing it systematically, and storing it securely to facilitate analysis that answers specific business or research questions (source content).
• Data-driven recommendations are derived from analyzing business data, providing actionable insights that help in decision-making and strategy formulation (source content).
• Population parameters are fixed values that describe the entire population but are often unknown; instead, analysts use sample statistics to estimate these parameters (source content).
• Samples are used because collecting data from the entire population is often impractical; they provide a manageable way to infer about the whole population with the understanding that sample statistics may differ from true population parameters.
• Proper organizing and storing of data ensures data integrity, accessibility, and readiness for analysis, which is crucial for producing reliable, data-driven recommendations (source content).

💡 Key Takeaway

The data collection and storage process is fundamental in data analytics, enabling accurate analysis and reliable data-driven recommendations by systematically managing data to estimate population characteristics through sample statistics.

📖 3. Descriptive Statistics

🔑 Key Concepts & Definitions

• Mean: Also known as the average, the mean is a measure of central tendency that sums all data points and divides by the total number of observations. (Source: "Mean: average representing data set")
• Median: The middle value in an ordered data set; if the data set has an odd number of observations, it is the central number; if even, it is the average of the two middle numbers. (Source: "Median: middle value when data sorted")
• Mode: The most frequently occurring value in a data set; suitable for qualitative data. (Source: "Mode: most frequent value")
• Dependent Variable (DV): The variable that is measured or observed to assess the effect of the independent variable; often the outcome of interest. (Source: "DV means follower I have from UK or not")
• Independent Variable (IV): The variable that is manipulated or categorized to observe its effect on the dependent variable; often the cause or input. (Source: "IV what my nationality")**
• Null Hypothesis (H0) and Alternative Hypothesis (H1): H0 states there is no effect or difference (e.g., no effect of feed type on chicken weight), while H1 suggests there is an effect (e.g., feed type affects chicken weight). (Source: "H0 Null, not effecting", "H1 Alternative, yes there is effect")

📝 Essential Points

• Descriptive statistics involve calculating measures like mean, median, and mode to summarize data and identify central tendency. The mean is sensitive to outliers, whereas the median provides a better central value when outliers are present.
• The mode is particularly useful for qualitative data or categorical variables, indicating the most common category or value.
• Understanding the relationship between variables involves hypotheses testing: H0 (null hypothesis) assumes no effect, while H1 (alternative hypothesis) indicates a potential effect, guiding statistical inference.
• Variance and standard deviation are measures of dispersion, showing how spread out data points are around the mean, which is crucial for understanding data variability.
• Sampling methods like simple random, stratified, systematic, and cluster sampling are techniques to select representative subsets of a population for analysis, reducing bias and improving inference accuracy.

💡 Key Takeaway

Descriptive statistics summarize and describe data through measures like mean, median, and mode, while hypotheses (H0 and H1) guide the interpretation of relationships between variables, providing foundational insights for further analysis.

📖 4. Predictive Analytics

🔑 Key Concepts & Definitions

Predictive analytics: Building models to predict numerical and categorical outcomes based on historical data, enabling forecasts of future events or behaviors.

Use of historical data for prediction: Leveraging past data to develop models that can estimate future results, assuming patterns in the data will continue over time.

Classification prediction: A type of predictive modeling that assigns data points to predefined categories or classes, such as spam detection or customer segmentation.

Regression prediction: A form of predictive modeling used to estimate continuous numerical outcomes, such as sales figures or temperature levels.

📝 Essential Points

Predictive analytics involves creating mathematical and statistical models that analyze past data to forecast future events or classify data into categories. It relies heavily on historical data, assuming that past patterns will persist. Classification models predict categorical outcomes, while regression models estimate numerical values. These models are essential for making data-driven decisions in various fields, from marketing to finance. The process requires selecting appropriate algorithms and validating the models to ensure accuracy and reliability.

💡 Key Takeaway

Predictive analytics uses historical data to build models that forecast future outcomes or classify data, playing a crucial role in data-driven decision-making by leveraging patterns in past information.

📖 5. Prescriptive Analytics

🔑 Key Concepts & Definitions

• Prescriptive analytics: Using data assets as a guide to control and change the outcome of an event by building data-driven solutions that influence future results. It involves recommending actions based on data analysis to achieve desired objectives.

• Building data-driven solutions: Developing strategies or systems that utilize analyzed data to make informed decisions and implement changes aimed at influencing specific events or outcomes.

📝 Essential Points

Prescriptive analytics goes beyond descriptive and predictive analytics by actively shaping future events through strategic interventions. It leverages insights from data to formulate actionable recommendations, effectively enabling organizations to control outcomes rather than just understand or forecast them. The process involves designing solutions that are guided by data, with the goal of influencing variables and achieving targeted results.

💡 Key Takeaway

Prescriptive analytics focuses on using data to actively control and modify future outcomes by creating solutions that influence events, making it a proactive approach to decision-making.

📖 6. Measures of Central Tendency

🔑 Key Concepts & Definitions

• Mean: The average value of a data set, calculated by summing all data points and dividing by the number of points. It represents a central value that best summarizes the data (see source content: "Mean also called average, tries to find out the number which best represent a list of number").
• Median: The middle value when data points are arranged in ascending or descending order. If the data set has an odd number of observations, the median is the central number; if even, it is the average of the two middle numbers. It is less affected by outliers (see source content: "Arrange the numbers from smallest to largest number. If the number is ODD: Single number exactly in the middle").
• Mode: The most frequently occurring value in a data set. It is particularly suitable for qualitative data and categorical variables (see source content: "Mode is frequent one, suitable for find the height of distribution").

📝 Essential Points

• The mean is sensitive to outliers, which can skew the average, especially if a dataset contains extreme values.
• The median provides a better measure of central tendency in skewed distributions or when outliers are present, as it is not influenced by extreme values.
• The mode is useful for identifying the most common category or value, especially in non-numeric or categorical data. It is less affected by outliers compared to the mean.
• When data is symmetric and free of outliers, the mean and median tend to be similar. In skewed data, the median often provides a more accurate central location.
• These measures help in understanding the typical value in a dataset, guiding decision-making and data interpretation.

💡 Key Takeaway

The mean, median, and mode are fundamental measures of central tendency that describe the typical or central value of a dataset, each with strengths suited to different data types and distributions.

📖 7. Measures of Spread

🔑 Key Concepts & Definitions

• Variance: VARIANCE (see section 11): a statistical measure that quantifies how far numbers in a data set are spread out from their mean, calculated as the average of squared deviations from the mean. It provides an overall sense of data dispersion.

• Range: The difference between the maximum and minimum values in a data set. It offers a simple measure of spread, indicating the total extent of the data.

• Standard Deviation: STANDARD DEVIATION (see section 11): a measure that assesses the typical distance of data points from the mean, calculated as the square root of variance. It is widely used to understand data variability in units consistent with the original data.

📝 Essential Points

• Variance gives a numerical value representing the average squared deviation from the mean, helping to understand the overall spread of data points (see VARIANCE). It is particularly useful in statistical modeling and hypothesis testing.

• Range is a straightforward measure of spread, calculated as the difference between the highest and lowest data points, but it can be sensitive to outliers and does not provide information about data distribution within the range.

• Standard deviation is derived from variance and provides an intuitive measure of dispersion in the same units as the original data, making it easier to interpret in practical contexts (see STANDARD DEVIATION).

• These measures are essential in descriptive analytics for understanding the variability and consistency of data, which can influence decision-making and further statistical analysis.

💡 Key Takeaway

Variance, range, and standard deviation are fundamental measures of spread that quantify how data points are dispersed around the central tendency, with variance and standard deviation offering detailed insights into data variability beyond simple range.

📖 8. Sampling Methods

🔑 Key Concepts & Definitions

• Sampling: The process of selecting a subset from a population to make inferences about the entire population (source content). It allows for analysis without examining every individual.

• Simple Random Sampling: A sampling method where each individual in the population has an equal chance of being selected, ensuring unbiased representation (source content).

• Stratified Sampling: Dividing the population into distinct groups or strata based on specific characteristics (e.g., age, grade), then randomly sampling from each group to ensure proportional representation (source content).

• Systematic Sampling: Selecting every nth item from an ordered population list, such as every 3rd or 10th item, often used in industrial or manufacturing settings (source content).

• Cluster Sampling: Sampling based on geographic or proximity clusters, where entire groups or locations are randomly selected, such as choosing specific hospitals or boxes of apples (source content).

📝 Essential Points

• Sampling is a shortcut to infer characteristics of the entire population when full data collection is impractical.
• Simple random sampling ensures each individual has an equal probability, reducing selection bias.
• Stratified sampling helps prevent over- or under-representation of subgroups, especially when groups vary significantly in size or importance.
• Systematic sampling is useful in ordered settings, but care must be taken if the order introduces bias.
• Cluster sampling is efficient when populations are naturally divided into clusters, such as geographic regions, but may increase sampling error if clusters are heterogeneous.
• The choice of sampling method impacts the accuracy and representativeness of the results.

💡 Key Takeaway

Sampling methods—simple random, stratified, systematic, and cluster—are essential tools for efficiently and accurately estimating population characteristics, each suited to different research contexts and data structures.

📖 9. Types of Data

🔑 Key Concepts & Definitions

• Nominal data: Categorical data without any inherent order or ranking. Examples include eye color or gender. Suitable for mode analysis but not for calculating averages (source content).
• Ordinal data: Categorical data with a specific order or ranking but without consistent intervals between categories. Examples include GCSE, undergraduate, master, PhD. Can be ordered but not mathematically manipulated (source content).
• Discrete data: Quantitative data that consists of whole numbers only, with no intermediate values possible. Examples include the number of children or units produced. Cannot take fractional or decimal values (source content).
• Continuous data: Quantitative data that can take any value within a range, including decimals. Examples include height, weight, or temperature. Suitable for interval and ratio scales (source content).
• Interval scale: Numeric scale with equal intervals but no true zero point, meaning ratios are not meaningful (e.g., temperature in Celsius). Zero does not indicate absence of the quantity (source content).
• Ratio scale: Numeric scale with a meaningful zero point, allowing for ratio comparisons (e.g., weight, height). Zero indicates the absence of the measured quantity (source content).

📝 Essential Points

• Data types influence the choice of analysis methods: qualitative data (nominal and ordinal) are analyzed using mode and ranking, while quantitative data (discrete and continuous) are analyzed using measures like mean, variance, and standard deviation.
• Nominal data are categorical without order; ordinal data have a natural order but lack consistent intervals.
• Discrete data are whole numbers, suitable for counts; continuous data include any value, including decimals, allowing for more precise measurement.
• Interval and ratio scales are both numeric, but the key difference is the presence of a true zero in ratio scale, enabling meaningful ratio comparisons.
• Understanding data types is crucial for selecting appropriate descriptive statistics and analysis techniques (source content).

💡 Key Takeaway

Data types—nominal, ordinal, discrete, and continuous—determine the appropriate methods for analysis and interpretation, with ratio and interval scales providing the foundation for numeric measurement.

📖 10. Data Measurement Scales

🔑 Key Concepts & Definitions

• Interval scale: A numeric scale with equal intervals between values but no true zero point, meaning the zero does not indicate the absence of the quantity being measured. For example, temperature in Celsius or Fahrenheit, where 0 does not mean 'no temperature'. (source content)

• Ratio scale: A numeric scale with equal intervals and a meaningful zero point, allowing for ratio comparisons. For example, weight or height, where zero indicates the absence of the measured attribute, and ratios like "twice as much" are meaningful. (source content)

📝 Essential Points

• The interval scale allows for the measurement of differences between data points but does not support ratio calculations because there is no true zero (e.g., 20°C is not twice as hot as 10°C). It is useful for measuring temperature, dates, or other quantities where the zero point is arbitrary.
• The ratio scale supports all mathematical operations, including ratios, because it has a true zero point (e.g., weight, height). This makes it suitable for measurements where the absence of the attribute is meaningful and ratios are interpretable.
• Understanding the difference between these scales is crucial for selecting appropriate statistical methods and interpreting data correctly, especially when calculating measures like mean, median, or standard deviation.
• The source emphasizes that interval scales do not permit meaningful ratio comparisons, unlike ratio scales, which do.

💡 Key Takeaway

Interval scales measure data with equal intervals but lack a true zero, limiting ratio comparisons, whereas ratio scales have a meaningful zero point, allowing for comprehensive ratio-based analysis.

📖 11. Dispersion Measures

🔑 Key Concepts & Definitions

• Dispersion: Variability in data beyond central tendency, indicating how spread out the data points are around the central value (see source content).
• Variance: A measure of dispersion that quantifies how far numbers in a data set are spread from the mean, calculated as the average of squared deviations from the mean (source content).
• Standard Deviation: The square root of variance, providing a measure of the average distance of data points from the mean, useful for assessing data variability (source content).
• Range: The simplest measure of spread, calculated as the difference between the maximum and minimum values in a data set, indicating the total spread of data points (source content).

📝 Essential Points

• Dispersion measures are essential because central tendency alone does not fully describe data variability, which impacts the interpretation of data analysis (source content).
• Variance and standard deviation are suitable for all numeric data types, providing detailed insights into data spread (source content).
• Range offers a quick, simple measure of spread but is sensitive to outliers, as it only considers the two extreme values (source content).
• Understanding dispersion helps in comparing datasets, such as assessing inequality or consistency, exemplified by comparing standard deviations in income levels across countries (source content).

💡 Key Takeaway

Dispersion measures like variance, standard deviation, and range provide critical insights into data variability beyond central tendency, enabling more comprehensive data analysis and comparison.

📖 12. Outliers and Influence

🔑 Key Concepts & Definitions

• Outliers: Extreme values in a data set that are significantly higher or lower than most other data points, which can skew the calculation of the mean (see measure of central tendency).
• Influence of Outliers on Mean vs Median: Outliers tend to have a strong effect on the mean, pulling it toward the extreme value, whereas the median remains relatively unaffected because it depends on the middle position of ordered data (see measure of central tendency).
• Mode as Less Affected by Outliers: The mode, being the most frequently occurring value, is generally less influenced by outliers, especially in qualitative data, making it a stable measure in the presence of extreme values.

📝 Essential Points

• Outliers can distort the mean, making it less representative of the typical data point, especially when the outlier is an extreme value (source content).
• The median is more robust against outliers because it depends on the middle position in the ordered data, not the magnitude of the values (source content).
• The mode is particularly useful for categorical or qualitative data, where the most common category is unaffected by outliers or extreme values (source content).
• Detecting outliers involves examining the data for unusually high or low values, which may be identified visually (e.g., box plots) or through statistical methods.
• Outliers can be genuine or due to errors; understanding their influence helps in choosing the appropriate measure of central tendency for analysis.

💡 Key Takeaway

Outliers are extreme values that can skew the mean, but the median and mode are less affected, making them more reliable measures of central tendency when data contains outliers.

📅 Key Dates

(OMITTED: No significant dates provided in the content)

📊 Synthesis Tables

⚠️ Common Pitfalls & Confusions

• Confusing correlation with causation in descriptive and predictive analytics.
• Overlooking the impact of outliers on measures like mean and standard deviation.
• Misinterpreting the null hypothesis (H0) as evidence of no effect without considering p-values.
• Assuming predictive models are infallible; neglecting model validation and overfitting.
• Using inappropriate sampling methods that introduce bias, affecting sample statistics.
• Confusing measures of central tendency (mean vs. median) in skewed distributions.
• Ignoring the scale of data when choosing measures of spread or dispersion.
• Misapplying data measurement scales, e.g., treating ordinal data as interval data.
• Overlooking the influence of outliers on measures of spread and central tendency.
• Misunderstanding the difference between population parameters and sample statistics.
• Assuming all data collection methods are equally valid without considering data quality and relevance.

✅ Exam Checklist

• Know the definitions and differences between descriptive, predictive, and prescriptive analytics.
• Understand the role of measures of central tendency: mean, median, and mode.
• Be able to explain measures of spread: variance and standard deviation.
• Recognize different sampling methods: simple random, stratified, systematic, cluster.
• Distinguish between types of data: nominal, ordinal, interval, ratio.
• Master data measurement scales and their appropriate analysis techniques.
• Understand how to identify and handle outliers and their influence on data analysis.
• Know the concepts of population parameters versus sample statistics.
• Be familiar with the use of hypotheses testing: null hypothesis (H0) and alternative hypothesis (H1).
• Understand the key concepts of regression and classification in predictive analytics.
• Know SMITH's definition of the invisible hand and its relevance to market equilibrium.
• Be able to describe the process of data collection, organization, and storage for analysis.
• Recognize the importance of data-driven recommendations in decision-making.