A Deep Dive into forecasting with Excel and Python.

Analyze time series with univariate statistics, assess mean, median, and outliers, apply confidence intervals and transformations, and explore correlations, autocorrelation, and seasonal lags to improve forecasting.

Plot the time series to reveal a seasonal pattern and rising trend in monthly Australian gas production, and compute univariate statistics including mean, variance, and standard deviation.

Explore univariate statistics for forecasting with Excel and Python, including the coefficient of variation, mean and median, outliers, and confidence intervals to assess data variability.

Explore bivariate statistics for time series by computing covariance and correlation between gas production (X) and time (Y), revealing a positive linear relationship and the role of scaling in correlation.

Develop an understanding of auto-correlation in time series by analyzing lagged relationships in seasonal gas production data, computing means and deviations, and interpreting nonstationarity using Excel and Python.

Explore the assignment on univariate and bivariate statistics using retailer sales and gas production datasets, compute mean, variance, standard deviation, coefficient of variation, outliers, log transform, correlations, and lag-two autocorrelation.

Explore univariate and bivariate analysis, compute variance and standard deviation, handle outliers, impute zeros for log transforms, and assess covariance, correlation, and lag-two autocorrelation for mild serial correlation.

Explore univariate statistics, correlation, covariance, and autocorrelation with practical Excel exercises, highlighting challenges and setting up the forecast package extensions for future sections.

Explore simple forecasting methods—from naive benchmarks to seasonal and moving averages—using linear regression and accuracy measures such as mean absolute error, mean absolute percentage error, and mean absolute scaled error.

Explore naive and seasonal naive forecasting as simple baseline methods, learn training-testing split, compute forecast errors, and assess bias with mean error and insights on overshooting and undershooting.

Explore forecast evaluation with Excel and Python by comparing naive and seasonal naive models using mean absolute error, mean squared error, and mean percentage error on gas production data.

Compare seasonal forecasting methods by computing errors such as mean absolute percentage error and mean absolute scaled error for the seasonal average versus seasonal naive, using in-sample and out-of-sample data.

Compute the mean absolute scaled error by dividing the MAE of the seasonal average on the testing set by the in-sample MAE of the seasonal naive, enabling method comparison.

Explore log transformations to stabilize variance and apply simple exponential smoothing with adjustable alpha, comparing logged versus non-logged forecasts.

Compare naive, seasonal naive, moving averages, weighted moving averages, simple exponential smoothing, and linear regression to forecast gas production and test accuracy on a testing dataset.

Explore naive, seasonal naive, simple average, and moving average forecasting methods, and learn how simple methods can sometimes outperform complex models on the test set.

Explore forecasting with Excel and Python by building linear regression models, custom weighted moving averages, and simple exponential smoothing to compare accuracy on training and test data.

Fit training data to three models - weighted moving average, custom weighted moving average, and simple exponential smoothing - and optimize weights to minimize mean squared error.

Compare forecasting methods using error metrics such as mean squared error, mean absolute error, mean absolute percentage error, and RMSE. Identify the customized weighted moving average as the winner.

Apply simple forecasting methods to retailer sales data, including exponential smoothing and log exponential smoothing, with seven-day seasonality and moving averages, using training/testing splits to identify the best approach.

compare forecasting methods in excel and python, from exponential smoothing to seasonal averages. identify seasonal average as the best simple forecast based on mae, mse, and rmse.

Explore excel-based forecasting and accuracy measures like mean absolute error, mean absolute percentage error, and mean squared error. Use linest and solver to optimize alpha in a weighted moving average.

Master time series analysis with moving averages and trend extraction, distinguish smoothing from forecasting, and apply additive or multiplicative decomposition for seasonal adjustment and forecasting.

Explore moving averages to smooth data, decompose time series into trend, seasonality, and randomness, and compare simple, double, and centered moving averages.

Learn how to extract and smooth the trend with moving averages in time series: simple, double, and centered moving averages, and distinguish trend, seasonality, and errors.

Learn time series decomposition in Excel, separating data into trend, seasonality, and error, and compare additive and multiplicative models. Understand when variance and log transformation dictate the model choice.

Explore additive time series decomposition by extracting the trend with a centered moving average, removing trend to reveal seasonality, and estimating seasonal indices and errors to validate the model.

Apply multiplicative decomposition to a gas production time series with rising variance, using division by trend, seasonal indices, and a multiplicative error to confirm seasonality and trend.

Practice solving seven-day seasonality decomposition to forecast with Excel and Python using moving averages (seven, double, centered) and analyzing additive and multiplicative time series to extract trend, seasonality, and errors.

Explore how to perform decomposition for forecasting using moving averages, centering choices, additive vs multiplicative models, remove trend, handle seasonality and outliers, and validate results.

Learn to detrender data, apply seasonal adjustments, and extract time-series components, using double or centered moving averages for smoothing, and choose multiplicative versus additive decomposition based on seasonal variance.

Explore exponential smoothing, including simple, halt trend, and triple methods with level, trend, and seasonality. Optimize alpha, beta, gamma in Excel and compare its accuracy with multiple linear regression.

Explore exponential smoothing concepts, including simple exponential smoothing and Holt and Holt-Winters models, using alpha, beta, and gamma to capture level, trend, and seasonality in demand forecasts.

Learn Holt exponential smoothing, an extension of simple exponential smoothing for data with trend, using alpha and beta to forecast future values in Excel and Python.

Explore simple exponential smoothing and Holt's method by initializing alpha and beta, applying forecast equations to gas production data, and comparing training and test horizons in Excel or Python.

Explore Holt's exponential smoothing in Excel by updating level and trend with alpha and beta to forecast tomorrow, and compare MSE, RMSE, and mean absolute percentage error.

Explore Holt-Winters exponential smoothing with additive and multiplicative models, detailing level, trend, and seasonality to forecast gas production, fit training data with solver optimization, and forecast 12 months ahead.

Apply an additive holt-winters exponential smoothing model, optimizing alpha, beta, and gamma to reduce training and test error. Produce a 12-month forecast from level, trend, and seasonality.

Explore multiplicative Holt-Winters forecasting in Excel and Python, building level, trend, and multiplicative seasonality using division and the yearly average, with a forecast aligned to last year's seasonality.

Optimize alpha, beta, gamma under constraints in multiplicative exponential smoothing to fit training and testing data, and forecast 12 months ahead with level, trend, and seasonality for daily data.

Apply exponential smoothing and Holt methods to UK retailer sales, compare forecasts using mean absolute error, and optimize alpha, beta, gamma for additive and multiplicative seasonal models to seven-day forecasts.

Analyze retail sales using exponential smoothing methods—simple, Holt, additive and multiplicative models; optimize parameters by MAE and select triple exponential smoothing with multiplicative seasonality as the best forecast.

Learn how multiple linear regression enhances forecasting by incorporating trend, seasonality, holidays, and promotions with dummy variables, and compare scenarios and p-values for feature significance.

This lecture introduces linear regression as a forecasting tool, showing how price, trend, and seasonality form a multiple regression model to predict demand from independent variables.

Apply multiple linear regression in Excel to gas production data using trend, seasonality, and lags, with monthly dummy variables, and run regression via the data analysis Toolpak.

Fit a multiple linear regression model for gas production, interpret R-squared and p-values, compare exponential smoothing methods, and forecast 12 months using feature engineering and dimensionality reduction.

Install and set up Python via Anaconda, explore Spyder and Jupyter notebooks, and learn Python basics, lists, NumPy arrays, dates, recency analysis, resampling, and forecasting with statsmodels.

Explore how Python evolved into a data science cornerstone, and learn to install Anaconda with Jupyter Notebook and Spyder, using the same code across interfaces.

Download the Anaconda individual edition and install it, starting with the Mac version. Learn to access the installers for Windows, Mac, or Linux and complete the setup.

Install the Anaconda package by following the on-screen prompts, then explore inside Anaconda to discover which applications will run.

Explore Spyder overview and its integration with Anaconda, Notebook, and Jupyter, and learn to set up data science projects, run code in the console, and manage scripts.

Learn how to run Python code in a Jupyter notebook within Anaconda. See instant outputs in Jupyter Lab and notebook, and compare Spyder workflows while ensuring identical code works.

Explore how Python libraries and packages, like Panda, NumPy, and Matplotlib, enable data manipulation, arrays, visualization, and forecasting within Jupyter Notebook or Spyder using Anaconda.

Install and configure Python tools like Spyder and Jupyter Notebook, and learn to update libraries with pip and conda. Practice Python fundamentals using Jupyter Notebook and engage in q&a.

Discover Python data types and structures, including numbers, strings, datetime, booleans, lists, arrays, dictionaries, and tuples, and learn to parse and clean comma-formatted numbers for analysis.

Explore Python by using parentheses for function calls, assigning variables with equals, and distinguishing strings from objects, while introducing data frames as Excel-like tables of observations (rows) and attributes (features).

Practice Python fundamentals in a Jupyter notebook by performing basic arithmetic, creating variables, and building lists with indexing that starts at zero.

Learn to create and manipulate lists in Python for forecasting with Excel and Python, understand zero-based indexing, and practice slicing to subset data with strings and name lists.

Explore dictionaries in Python by mapping keys to values, creating dictionaries with curly braces, and retrieving keys and values using dict.keys and dict.values via dot notation.

Learn how to use NumPy arrays for efficient data storage in forecasting with Python, compare arrays to lists, explore multidimensional arrays, and apply arithmetic operations.

Import the online retail data in Python using pandas read_csv in a Jupyter notebook, then explore the dataframe with head, tail, shape, and describe to reveal data characteristics.

Subset data frames with iloc and loc to select rows and columns by index or name, including the first five, and add a new column such as price in euros.

Explore testing values with conditions, including greater than, less than, and equal, using multiple criteria. Apply these in for loops and if statements for forecasting data with Excel and Python.

Write a Python function named status to classify a person as child, teenager, or adult using if/elif/else, handle indentation, and print or return results for a list of people.

Learn how to apply a function to every element in a list with Python's map, create a map object, and convert it to a list to display each element's status.

Apply for loops to perform iterations across a list, printing each age from a class list and obtain the same result as the map function.

Apply a status function to a list of ages with a for loop, storing results with append. Compare return, print, and map within forecasting with Excel and Python.

Define a function to check if a country equals United Kingdom in a retail data frame. Use a map to create a new boolean column indicating United Kingdom membership.

Apply a for loop to a data frame to process each row, create a new column, and compare to mapping, focusing on the first ten rows.

Master Python fundamentals for forecasting, including lists, dictionaries, and pandas, and learn to subset with iloc and loc, define functions, map categories, and loops.

Guide to solving a Python assignment on a cars dataset, exploring rows and columns, horsepower and price, and creating a pricing category with budget, suitable, and expensive labels.

Solve assignment by analyzing the cars dataset in pandas, focusing on horsepower, price, and highway mpg, using shape and describe to identify Porsche 911 GT2 as the most expensive.

Rename the dataset column to car name and create a car pricing subset from price. Implement pricing_category to label budget, suitable, or expensive cars, then apply it across rows.

Master date handling in Python and pandas for forecasting with time-based analyses. Parse dates with to_datetime, extract time components, and perform resampling and moving averages.

Clean retail data with pandas and numpy, drop duplicates and missing rows. Convert invoice date to datetime in Python, format with strftime, and extract components.

Compute customer recency by extracting the last purchase date per customer, subtracting from the dataset max date, and interpreting the distribution with a histogram using Excel and Python.

Convert time delta to days and extract its components to measure recency from the last purchase date, and analyze the recency histogram showing a right-skew toward recent purchases.

Learn to model inter-arrival times for customers using Excel and Python by extracting unique customers, creating a dataframe with each customer and date, and computing previous dates with per-customer shifts.

Model inter-arrival times per customer using python by grouping data, adding a previous date column, and concatenating per-customer results to avoid cross-customer shifts.

Convert date fields to date time, compute durations between consecutive purchases, and group by customer to derive the mean duration as the inter-arrival time.

Resample time series data in Python by converting date to a datetime index, then aggregate by month, week, or year using mean, sum, first, last, and std for visualization.

Explore rolling time series by computing moving averages with rolling windows, such as seven days, to create new data points from previous values, illustrated with MSFT stock data.

Explore rolling time series with weekly and monthly windows, using a 30-day window on msft to smooth data and reveal buy or sell signals via moving averages in finance.

Learn to extract date components with pd.to_datetime, apply resampling and rolling calculations, and use moving averages to forecast stock trends and support financial decisions.

Solve section seven assignment by cleaning the 2011 data subset, creating week day and month-year columns, validating dates, computing recency and last purchase, and applying moving averages with weekly resampling.

Preprocess 2011 retailer sales by converting invoice dates to datetime, extracting week, day, month, and year, then compute recency and apply moving averages and weekly resampling for forecasting.

Explore statistical forecasting with ARIMA and exponential smoothing, performing hyperparameter tuning on the three parameters and their orders to optimize forecasts using Excel and Python.

Explore ARIMA and exponential smoothing methods for time series forecasting, including autoregressive, integrated, and moving average components, differencing for stationarity, and trend and seasonality extraction for demand planning.

Compare four popular forecast accuracy measures—mean absolute error, mean squared error, root mean square error, and mean error (bias)—to understand overshoot and undershoot.

Explore time series forecasting with ARIMA and the Facebook Prophet library, using monthly revenue data and treating the time series as regressors, while examining seasonality and trend.

Explore time-based components, such as trend, seasonality, and remainder, and regression-based components for ARIMA, using Python's seasonal_decompose to extract trend, seasonality, and residuals, and measure their strengths.

Explore time series components by decomposing trend, seasonality, and remainder using the statsmodels time series analysis module. Learn how seasonal decomposition supports additive or multiplicative models for monthly series.

Decompose a monthly time series into trend, seasonality, and residuals, plot the components, and tailor forecasts by isolating seasonality or trend for clearer insights.

Explore arima modeling by regressing current observations on past values and past errors, using autoregression (p), integration (d), and moving average (q) with acf and pacf tools.

Assess stationarity in Python using rolling mean and standard deviation. Use the Fuller test and ARIMA integration to confirm data is stationary for forecasting.

Determine the best model for a stationary monthly time series by examining ACF and PACF, testing AR, MA, ARMA, and ARIMA with Statsmodels.

Assess ARIMA diagnostics by examining residuals, histograms, Q-Q plots, and correlograms to identify seasonality and distribution deviations, improving model selection beyond naive ARIMA before final seasonal components.

Perform a grid search over ARIMA and seasonal components using itertools to find the best fit for a monthly series with seasonality 12, stabilizing residuals.

We loop through ARIMA model combinations, store P, D, Q and seasonal orders in a data frame, fit models, and continue after errors with try/except.

Learn robust error handling in forecasting workflows by using Python try/except, skipping failing iterations, and selecting the best model based on AIC.

Fit the best model to our data, examine the fitting results and predicted mean, and forecast monthly observations for up to 12 steps using the Excel and Python workflow.

Explore forecasting with Excel and Python by computing and visualizing the forecast mean against actual values, and evaluate accuracy using mean absolute error.

Compare the ARIMA model with the intuitive forecast, fit the models in Excel and Python, and evaluate predictions using mean absolute error to balance complexity and accuracy.

Explore forecasting techniques for time series analysis, including single exponential smoothing with alpha, Holt’s method for trend, and Holt-Winters with additive or multiplicative seasonality.

Learn to apply Holt-Winters exponential smoothing in python, testing four configurations—additive and multiplicative with trend and seasonality—using a seasonality period of 12 and comparing models by AIC, BIC, and coefficients.

Compare four models using mean absolute error to select the best exponential smoothing or ARIMA approach for forecasting monthly revenue of the UK retailer; forecast 12 months and plot results.

Explore time series concepts including stationarity, ARIMA and exponential smoothing for daily, weekly, and monthly data. Understand ARIMA components, seasonal ARIMA, and forecasting segmentation for product segmentation.

Explore time series forecasting by analyzing retailer sales, cleaning and resampling to monthly data, and comparing seasonal ARIMA and exponential smoothing models using grid search, AIC, and 12-month forecasts.

Explore forecasting with Python by building a monthly time series from retail data, performing seasonal decomposition, and assessing stationarity to interpret trend, seasonality, and residuals.

Apply the Fuller test to confirm stationarity of the monthly series, examine the ACF and PACF, compare MA, AR, and ARIMA models, and explore seasonal p,d,q combos using AIC.

Examine multiple ARIMA models for forecasting with seasonality; compare differencing of one year vs two years, finding that one year differencing yields better mean absolute error and a reliable forecast.

Compare forecast models using mean absolute error and root mean squared error to select the additive Holt-Winters exponential smoothing. Forecast 12 months ahead for supply chain applications.