OpenEDG PCAD key terms

A database object that keeps track of a position in a result set and lets Python fetch rows one at a time or in batches after a query executes.

The Python handle to a database session; it opens the channel used to run statements, manage transactions, and produce cursors.

A SQL statement in which placeholders are filled with values supplied separately by the driver, preventing SQL injection and handling quoting automatically.

An attack in which untrusted input is concatenated into SQL so it alters the statement's logic; parameterized queries and input validation are the standard defenses.

The practice of retrieving a large API result set in successive pages using limit, offset, or cursor tokens until no further records are returned.

A server-side cap on how many requests a client may send in a time window; clients must back off and retry rather than hammering the endpoint.

Converting an API's JSON text into Python objects with the json module, then navigating nested dictionaries and lists to extract the needed fields.

Programmatically downloading web pages and extracting structured data from their HTML, typically with a parser such as BeautifulSoup rather than fragile text splitting.

Checking a site's robots.txt file before automated collection and honoring its disallow rules, along with terms of service and copyright constraints.

An absent or unrecorded observation, represented in pandas as NaN or None; it must be detected and handled explicitly because it silently breaks many computations.

Filling missing values with a substitute such as the mean, median, mode, or a model-based prediction, chosen so the substitution does not distort later analysis.

An observation far from the bulk of the data; it may be a genuine extreme case or a recording error, so it must be investigated rather than deleted automatically.

Converting a column to the correct type, such as text digits to integers or strings to datetimes, so that arithmetic, sorting, and joins behave correctly.

An explicit constraint a dataset must satisfy, such as a range check, uniqueness check, or referential check, used to detect bad records before analysis.

A row that repeats another row's identifying key, often introduced by merges or repeated loads; detection requires deciding which columns define identity.

A pandas two-dimensional table with labeled rows and columns, supporting vectorized operations, grouping, joins, and mixed column types.

A pandas one-dimensional labeled array; each column of a DataFrame is a Series, and most aggregation and vectorized operations return one.

Selecting rows by a condition that yields True or False per row, then using that array to filter the DataFrame; multiple conditions combine with & and |.

Splitting rows into groups by key values, applying a summary function such as mean or count to each group, and combining the results into a new table.

A reshaped summary that places one variable on rows, another on columns, and aggregated values in cells, useful for comparing groups across two dimensions.

Rules that let arrays of different but compatible shapes be combined in arithmetic without explicit copying, aligning dimensions from the right.

Applying an operation to a whole array at once in compiled code instead of looping in Python, which is faster and usually clearer.

A clause combining rows from two tables on a matching condition; inner, left, right, and full joins differ in which unmatched rows are kept.

A column or column set that uniquely identifies each row in a table and cannot be null; it enforces entity integrity.

A column referencing another table's primary key, enforcing referential integrity so that child rows cannot point at nonexistent parents.

GROUP BY collapses rows into groups for aggregation, while HAVING filters those groups after aggregation, unlike WHERE which filters individual rows first.

A transaction groups statements so they all succeed or all fail; commit makes changes permanent and rollback undoes uncommitted work after an error.

The mean is the arithmetic average and is sensitive to extreme values; the median is the middle value and resists them, so they diverge in skewed data.

The typical distance of observations from the mean, in the original units; a large value indicates high spread around the center.

The range between the 25th and 75th percentiles, covering the middle half of the data; it is a spread measure unaffected by extreme values.

The default claim of no effect or no difference that a significance test tries to reject using observed sample evidence.

The probability of observing data at least as extreme as the sample if the null hypothesis were true; it is not the probability that the null is true.

A range built from sample data that, under repeated sampling, would contain the true parameter in a stated percentage of cases.

Rejecting a true null hypothesis, a false positive whose rate equals the chosen significance level alpha.

The standard deviation of a statistic's sampling distribution; it shrinks as sample size grows and underlies confidence intervals and tests.

A standardized number between minus one and one describing the strength and direction of a linear association; it does not by itself establish causation.

A model that predicts a numeric target as a weighted sum of input features plus an intercept, fitted by minimizing squared residuals.

Dividing labeled data into a fitting portion and a held-out portion so performance is measured on examples the model never saw.

Fitting training data so closely, including its noise, that the model generalizes poorly to new observations; detected by a large train-versus-test gap.

Repeatedly splitting data into folds, training on some and validating on the rest, then averaging scores for a more stable performance estimate.

The proportion of variance in the target explained by the model; it rises with added features, so adjusted R-squared is preferred when comparing models.

The difference between an observed value and the model's prediction; residual plots reveal non-linearity, unequal variance, and influential points.

Matplotlib's two core objects: the Figure is the whole canvas, and each Axes is a plotting area with its own data coordinates, labels, and title.

Matching a visual form to the question: bars for category comparison, lines for trends over time, scatter for relationships, histograms for distributions.

The share of a graphic's ink that encodes actual data; maximizing it means removing gridlines, 3D effects, and decoration that add no information.

Starting a bar chart's value axis above zero, which exaggerates differences; bars must start at zero, though line charts may not need to.

A set of colors distinguishable under common color vision deficiencies, avoiding reliance on red versus green alone to encode categories.

An object used with the with statement that guarantees setup and cleanup, such as closing a file or database connection even when an exception occurs.

Catching errors with try, except, else, and finally so a program can recover, log the problem, or clean up instead of crashing abruptly.

A user-defined exception class, usually subclassing Exception, that names a specific failure so callers can catch it precisely.

A file, socket, or connection left open because cleanup was skipped, exhausting handles over time; context managers are the usual prevention.

Bundling data and the methods that operate on it inside a class and exposing only a controlled interface, often via properties and underscore-prefixed attributes.

A subclass reuses and extends a parent class, and objects of different classes can be used through the same interface when they share method names.

Collecting and using personal data only after telling people what is gathered, why, and with whom it is shared, and honoring their choice.

Removing or transforming identifiers so individuals cannot reasonably be re-identified, being stricter than pseudonymization which keeps a reversible mapping.

Collecting and retaining only the data actually needed for a stated purpose, reducing breach impact and storage obligations.

Systematic distortion from a sample that is not representative of the population, such as surveying only active users of a product.