## Basics ## Object Data - Row oriented data are known as Object Data. - Each Row -> Objects - Columns -> Attributes about objects (descriptors) #### Limitations of Object Data: ![[PMing_Analytics_CSV File.png]] - How I see is that the Object Data is 2 dimensional. It can just clarify what is visible. - In order to extract insights out of the data, we need to add depth to it and make it 3D. - Every change in an Object Data requires a lot of work, counting is inefficient. - It doesn't allow to do quick and efficient aggregations. ## Event Data ![[PMing_Analytics_Event Data 1.png]] - Instead of discrete Rows and Columns, we have **Keys** (name, date, age, favorite sandwich ) and **Values** (ak, 2021-04-20, 34, tuna salad) in Event Data, which come in pairs. - Data with this structure is called event data because it represents some event occurring. In this case, a person telling us there favorite sandwich. - Each occurrence (event) can have its own set of information that is completely independent from all the occurrences that have come before it and that will come after it. - The event data does not need to conform into arbitrary rows; columns. ## Advantages of Event Data ![[PMing_Analytics_Event Data.png]] - Event Data is really good for representing behaviours. - It should be used to record everything our end users do, inside our application. - Every Page they view - Every button they click - Every action they take - That's an event... ## Combining Event Data and Object Data - The event data and object data when combined, can be used to holistically analyse how users interact with our product. ![[PMing_Analytics_Event Data and Object Data.png]] - The object (user with keys and values) describes the user which has the name, email and favourite sandwich. - The Event represents a behaviour which has keys and values that describe that behaviour. **Product Analytics**: We want to analyse people ; given what we know about them ; and add in the context of what they have done. ## Event and Object Taxonomy - Taxonomy is a fancy way of saying: - What data do we have? What actions or attributes does the data represent? - This list of behaviours is called an Event Taxonomy. - In Mixpanel, you can add contextual information to the query by breaking the metric down using different properties (eg. Event Name) - Event names are same across all of the query builders within mixpanel. - Maintaining consistency with the event names is extremely important. - In order to Modify the events and object data, we use "**Lexicon**". Its a data dictionary. It comes in handy when we need to clean up the data or properly name the events. It allows you to add and update event property and descriptions, as well as other useful metadata. > Review both the Event and Object taxonomy To make sure the taxonomy is logical, and that each attribute's key clearly communicates what the underlying values in the data represent. - **EVENTS** | **PROPERTIES** | **OBJECTS** #### Questions: 1. In the context of object and event data, what is a taxonomy? A list of all keys (attributes) present in a set of data 2. What's the first step to create a clear data taxonomy? Ensuring all event and object keys (attributes) adhere to a consistent and legible naming convention 3. Which of the following is the best description for an event named "rate sandwich"? "Users may only rate sandwiches after seeing them; ratings are given in stars 1-5"