nicholasbierman / instantsearch-app

React Instantsearch App for NBA Teams & Players

Live link

Index Selection

The goal of this application is to create a search experience where users can search for NBA teams and players separately. I used two different indices from the provided `basketball` data set, one for `nba-teams` and `nba-players`.

Keeping separate indices, as opposed to having records for both teams and players in the same index, allowed me to tailor each index's configuration to its specific content while maintaining a flattened data structure.

From an implementation perspective, I indexed the data using the Algolia JavaScript API (scripts to configure each index can be found in the algolia directory) and built the search UI with the React InstantSearch library, Redux for global state management, and AWS S3 for image hosting. I chose this approach to illustrate my proficiency with these tools.

Searchable Attributes

• nba-teams: ["name"]
• nba-players: ["unordered(name)"]

In an effort to have as few searchable attributes as possible, I only set the `name` property to be searchable on each index. On the `nba-players` index, I used `["unordered(name)"]` to equally weight a player's first and last name at query time.

Filtering data

• nba-teams: ["location", "score"]
• nba-players: ["team", "points"]

While I did not want to include these properties in searchableAttributes, I improved the search experience and let users fine-tune their query by including them in attributesForFaceting at indexing time. I chose not to use the filterOnly method because I needed to include counts of each facet value on the frontend with the `RefinementList` widget.

Custom Ranking

• nba-teams: ["desc(score)"]
• nba-players: ["desc(points)"]

By default, when there is a tie between records in terms of textual relevance, the Algolia engine will use the objectID in alphanumeric order. Since a tie in terms of textual relevance is a high probability on each index (players with same first/last name on different teams, teams with same name in different locations), I set the `customRanking` attribute on each index at indexing time. In doing so, I created a search experience where the players with the most points or the teams with the highest score appeared earlier, allowing more relevant results in this use case.