datagovindia_vignette.Rmd

---
title: "Getting Started with datagovindia"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Getting Started with datagovindia}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```


**datagovindia** is a wrapper for around >130,000 APIS of the Government of India's open
data platform [data.gov.in](https://data.gov.in/ogpl_apis). Here is a small guide
to take you through the package.  Primarily,the functionality is centered around
three aspects :  

* **API discovery** - Finding the right API from all the available APIs
* **API information** - Getting information about a particular API 
* **Querying the API** - Getting a tidy data set from the chosen API

## Setup 
```{r setup}
library(datagovindia)
```


## API Discovery
The APIs from the portal are scraped every week to update a list of all APIs and 
the information attached to them like sector, source, field names etc. The website
[data.gov.in](https://data.gov.in/ogpl_apis) provides a search functionality through
string searches and drop down menus but these are very limited. The functions
in this package allows one to have more robust string based searches.  
A user can search by API title, description, organization type, organization (ministry),
sector and sources. Briefly there are two types of functions here, the first lets the 
user get a list of all available and unique organization type, organization (ministry),
sector and sources and the other lets one "search" by these criteria and more.


Here is a demonstration of the former (getting only the first few values)  


```{r}
###List of organizations (or ministries)
get_list_of_organizations() %>% 
  head

```

```{r}
###List of sectors 
get_list_of_sectors() %>% 
  head
```


### Searching for the right API

Once you have an idea about what you want to look for in the API, search queries 
can be constructed using titles, descriptions as well as the categories explored 
earlier. A data.frame with information of APIs matching the search keywords is 
returned. Multiple search functions can be applied over each other utilizing the
data.frame structure of the result. 


```{r,results='hide'}
##Single Criteria
search_api_by_title(title_contains = "pollution") %>% head(2)

```


```{r,echo=FALSE}
## Signle criteria
search_api_by_title(title_contains = "pollution") %>%
  head(2) %>% 
  knitr::kable()
```



```{r,results='hide'}
##Multiple Criteria
dplyr::intersect(search_api_by_title(title_contains = "pollution"),
                 search_api_by_organization(organization_name_contains = "pollution"))

```


```{r,echo=FALSE}
## Signle criteria
dplyr::intersect(search_api_by_title(title_contains = "pollution"),
                 search_api_by_organization(organization_name_contains = "pollution")) %>% 
  knitr::kable()
```

Once you have found the right API for your use, take a note of the "index_name"
of that API, for example, "0579cf1f-7e3b-4b15-b29a-87cf7b7c7a08" corresponds to
the API for "Details of Comprehensive Environmental Pollution Index (CEPI) Scores and Status of Moratorium in Critically Polluted Areas (CPAs) in India". **index_name** 
will be essential for both getting to know more about the API or to even get data from
it. 


## Getting more information about a chosen API  

There are two functions in this section, one to get API information, the other to get 
a available "field" names and types of the chosen API (using it's **index_name** obtained above). 


### API information

```{r,results='hide'}
get_api_info(api_index = "0579cf1f-7e3b-4b15-b29a-87cf7b7c7a08")
```

```{r,echo=FALSE}
get_api_info(api_index = "0579cf1f-7e3b-4b15-b29a-87cf7b7c7a08") %>% 
  knitr::kable()
```


### API Fields

Fields are essentially the variables in the dataset obtained from the API. Knowing
the fields before querying for the data will be essential to preform tasks such as filtering, sorting and subsetting the data obtained from the API's server. 

```{r,results='hide'}
get_api_fields(api_index = "0579cf1f-7e3b-4b15-b29a-87cf7b7c7a08")
```

```{r,echo=FALSE}
get_api_fields(api_index = "0579cf1f-7e3b-4b15-b29a-87cf7b7c7a08") %>% 
  knitr::kable()
```

The **id** of these fields is going to be useful while querying the data.

## Querying the chosen API

The function *get_api_data* is really the powerhouse in this package which allows 
one to do things over and above a manually constructed API query can do by utilizing
the data.frame structure of the underlying data. It allows the user to filter, sort, 
select variables and to decide how much of the data to extract. The website can itself 
filter on only one field with one value at a time but one command through the wrapper
can make multiple requests and append the results from these requests at the same time.  

But before we dive into data extraction, we first need to validate our API key relieved 
from [data.gov.in](https://data.gov.in/ogpl_apis). To get the key, you need to register first register and then get the key from your "My Account" page after logging in.
More instruction can be found on this [official guide](https://data.gov.in/help/how-use-datasets-apis). Once you get your API key, you 
can validate it as follows (only need to do this once per session) :

```{r}
##Using a sample key
register_api_key("579b464db66ec23bdd000001cdd3946e44ce4aad7209ff7b23ac571b")

```

Once you have your key registered, you are ready to extract data from a chosen API.
Here is what each argument means :  

* api_index : index_name of the chosen API (found by using search functions)
* results_per_req : Results per request sent to the server ; can take integer values or the string "all" to get all of the available data
* filter_by : A named character vector of field id (not the name) - value(s) pairs ; can take multiple fields as well as multiple comma separated values
* field_select : A character vector of fields to select only a subset of variables in the final data.frame
* sort_by : Sort by one or multiple fields


To recap, first find the API you want using the search functions, get the **index_name** of the API from the results, optionally take a look at the fields present in the data of the API and then use the get_api_data function to extract the data. 
Suppose we choose the API "Real time Air Quality Index from various location" with index_ name *3b01bcb8-0b14-4abf-b6f2-c1bfd384ba69*. First we will look at which fields are available to construct the right query.  
Suppose We want to get the data from only 2 cities Chandigarh and Gurugram and pollutants PM10 and NO2. We will let all fields to be returned (dataset columns).

We will use a sample key from the website for this demonstration. 

```{r}
register_api_key("579b464db66ec23bdd0000019fc84f43ca52437351b43702f5998234")
```
We now look at the fields available to play with. 

```{r,results="hide"}
get_api_fields("3b01bcb8-0b14-4abf-b6f2-c1bfd384ba69")
```

```{r,echo=FALSE}
get_api_fields("3b01bcb8-0b14-4abf-b6f2-c1bfd384ba69") %>% 
  knitr::kable()
```

We accordingly select the **city** and **pollution_id** fields for constructing our query. 
Note that we use only field id to finally query the data.

```{r,results='hide'}

get_api_data(api_index="3b01bcb8-0b14-4abf-b6f2-c1bfd384ba69",
             results_per_req=10,filter_by=c(city="Gurugram,Chandigarh",
                                            polutant_id="PM10,NO2"),
             field_select=c(),
             sort_by=c('state','city'))
```

```{r,echo=FALSE,message=FALSE}

get_api_data(api_index="3b01bcb8-0b14-4abf-b6f2-c1bfd384ba69",
             results_per_req=10,filter_by=c(city="Gurugram,Chandigarh",                                 pollutant_id="PM10,NO2"),
             field_select=c(),
             sort_by=c('state','district','city')) %>% 
  knitr::kable()
```