Hi. In the next set of videos, we will be discussing a commonly used and relatively sophisticated technique used often in marketing called conjoint analysis. So in this video I will talk about what is conjoint analysis and then in subsequent videos I'll talk about data collection methods, part-worth utilities, selecting attributes, how to actually perform the analysis in R, and how to interpret the results. While this term might sound a little complicated at first, conjoint, what does that mean? It's actually a very simple technique and it leverages some basic tools you already have in your toolkit, specifically, linear regression. So let's get started. So first, I want to describe what is conjoint analysis. First and foremost, it is a survey-based statistical technique and it's often used in marketing research or analytics. So the primary source of data is surveys. You will go out into the field and ask people, you and your sample, a bunch of questions, and from that you're going to try to identify their preferences. This technique helps to find the optimal combination of features in a product or service. So for example, so you have two sneakers, one is really high price and high-quality, one is low priced and maybe of a lower quality, and then one shoe is medium price but it's really good for running, for example. You would take these combinations of shoes and try to figure out which attribute is the most important. Is it the quality of the shoe, is it the price of the shoe or is it what the shoe is designed for running or some other activity, and you try to break out the weights that the people have in their heads. What is the relative value of price versus quality, price versus the type of use of the shoe, the quality versus the type of the use of the shoe? Try to break all those little components out to identify what's the most popular shoe. So some key things to know is that conjoint analysis is made up of factors and levels, and the factors are the variables that affect the purchase and the levels are the values assigned to each factor. So in the sneaker example that I just discussed, there might be a factor, price would be a factor for example, and what are the various levels? You can have a high price, medium price, low price, and you could even have a little more granular level data. Then so conjoint analysis tries to measure the effects of the levels of these factors by mixing them up. In conjoint analysis, you describe the features that you're interested in, and are meaningful to the respondents, and then ask them to rate the relative importance of these combinations. So it's a survey based technique. You're going to come up with their features, make a survey, and then ask them the questions. It can be considered, especially in the form that we're doing, an extension of multiple regression analysis and in our conjoint analysis we'll be using multiple regression as the engine. But note that conjoint analysis is really not tied to a specific statistical method. It is an overall problem set of trying to get an understanding of how people make decisions based on different attributes of a product. In this class, we will be discussing conjoint analysis using multiple regression, but note there are other techniques out there that are slightly more sophisticated. Also know that this is a baseline class, so once you understand these concepts, moving on to advanced concepts later on in your career should be a smooth transition. So using the multiple regression analysis, we try to find the best weights or combinations of the variables to produce the outcome, desired outcome that we're looking for. Underlying all these is a basic assumption that the buyers look at products as composed of various factors and they're weighing the factors and the levels in making their purchase decision. I'm sure if you reflect on your own purchasing decision, especially if you think about a big purchase where you have to really think about it, like buying a car or a new bicycle or something like that, you will weigh all the features and try to figure out what combination is the one you want at the right price point. So let me before I dive in talk about the vocabulary, the definitions that we use in conjoint analysis. First is attribute and factor. These are the subjective assessments of a product. They can be things like the price, the weight, the color, the quality, the usage, things like that. Then level is within the factor and that's how you measure each of these attributes or factors. High or low, is the price high or low? Is the weight heavy or light? Color, is it red or blue? Usage, is it for running or hiking, things like that? The next term that I'd like to describe is known as part-worth utilities. That is the extent to which a factor contributes to the whole utility of the product. So when I buy, for example, a new car, I value the comfort, I might value the handling, the safety features, and the price among other different types of attributes. Then what conjoint analysis does is try to figure out how much does price weigh into my decision-making process or how much does safety weigh into my decision-making process? There are a notion known as prohibited pairs and these are two factors that should never appear together. For example, a six-inch Apple phone for a $100. So these are "outside the realm of normal day-to-day possibilities." This is not something you would realistically consider. Buying a six-inch Apple phone for a $100. The cost far outweighs the price so that would really never happen in real life. So that wraps up the introduction to conjoint analysis and let's dive in in the next set of videos.