Prerequisites for Sentiment Analysis App using Swift<\/h3>\n
To follow along with this project, you will need the following:<\/p>\n
- \n
- Xcode 12 or higher.<\/li>\n
- Basic knowledge of SwiftUI and Core ML.<\/li>\n
- An Apple Developer account (to deploy the app to a device).<\/li>\n
- Dataset for Training the ML Model. It can be downloaded from here.<\/li>\n<\/ul>\n
Download Swift Sentiment Analysis App Project<\/h3>\n
Please download the source code of the Swift Sentiment Analysis App Project from the following link: Swift<\/strong> Sen<\/strong>timent Analysis App Project Code<\/strong><\/a>.<\/p>\n
Steps to Create Sentiment Analysis App using Swift<\/h3>\n
Following are the steps for developing the Swift Sentiment Analysis App Project:<\/p>\n
Step 1:<\/strong> Create a Sentiment Analysis model using Xcode playground and dataset.
\nStep 2:<\/strong> Create a new SwiftUI project in Xcode.
\nStep 3:<\/strong> Import the ML model into the project
\nStep 4:<\/strong> Create the app UI and functionality
\nStep 5:<\/strong> Test the app on a simulator or a physical iOS device.<\/p>\nStep 1: Create a Sentiment Analysis model using Xcode playground and dataset.<\/h4>\n
a.<\/strong> Open Xcode and select “Create a new Xcode playground.”<\/p>\n
![\"create](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/create-new-xcode-1-scaled.webp\")
<\/a><\/p>\n![\"xcode](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/xcode-playground.webp\")
<\/a><\/p>\nb.<\/strong> Choose a name and a location to save the playground<\/p>\n
![\"location](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/location-to-save-playground.webp\")
<\/a><\/p>\nNow you will see something like this<\/p>\n
![\"swift](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/swift-sentiment-analysis-output-scaled.webp\")
<\/a><\/p>\nc.<\/strong> In the first cell, import the necessary libraries:<\/p>\n
\/\/Importing the required libraries\r\nimport Cocoa\r\nimport CreateML\r\nimport Foundation\r\n<\/pre>\n
d.<\/strong> Provide the path for the dataset and output model.<\/p>\n
\/\/ Setting the input and output file paths\r\nlet inputFilePath = \"\/Users\/amanbind\/Desktop\/Model\/tweet2.csv\"\r\nlet outputFilePath = \"\/Users\/amanbind\/Desktop\/Model\/SentimentAnalysisClassifier.mlmodel\"\r\n<\/pre>\n
e.<\/strong> Load the data into a DataFrame and split it into training and testing data<\/p>\n
\/\/ Reading the data from the csv file\r\nlet data = try MLDataTable(contentsOf: URL(fileURLWithPath: inputFilePath))\r\n\r\n\r\n\/\/ Splitting the data into training and testing data\r\nlet (trainingData, testingData) = data.randomSplit(by: 0.8, seed: 5)\r\n<\/pre>\n
f.<\/strong> In the third cell, create a sentiment analysis model using the data and set the metadata of the model.<\/p>\n
\/\/ Creating the model\r\nlet sentimentClassifier = try MLTextClassifier(trainingData: trainingData, textColumn: \"Text\", labelColumn: \"sentiment\")\r\n\r\n\r\n\/\/ Setting the model's metadata\r\nlet metadata = MLModelMetadata(author: \"TechVidvan\", shortDescription: \"A model trained to classify the emotion of a sentence\", version: \"1.0\")\r\n<\/pre>\n
g.<\/strong> Testing the accuracy of the trained model<\/p>\n
\/\/ Testing the model\r\nlet evaluationMetrics = sentimentClassifier.evaluation(on: testingData, textColumn: \"Text\", labelColumn: \"sentiment\")\r\nlet evaluationAccuracy = (1.0 - evaluationMetrics.classificationError) * 100\r\n<\/pre>\n
h.<\/strong> Save the model to your local machine:<\/p>\n
\/\/ Saving the model\r\ntry sentimentClassifier.write(to: URL(fileURLWithPath: outputFilePath), metadata: metadata)\r\n<\/pre>\n
i.<\/strong> Run the playground file. It will train the model and generate the ML model file in the specified folder<\/p>\n
![\"run](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/run-the-playground-file-1-scaled.webp\")
<\/a><\/p>\nThe training accuracy of our model after 16 training iterations is 99.93% and our Testing accuracy is 82.0%<\/p>\n
![\"sentiment](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/sentiment-analysis-output.webp\")
<\/a><\/p>\nStep 2: Create a new SwiftUI project in Xcode.<\/h4>\n
a.<\/strong> Open Xcode and Click on the \u201cCreate a new Xcode Project\u201d option.<\/p>\n
![\"welcome](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/welcome-to-xcode.webp\")
<\/a><\/p>\nb.<\/strong> Now select platform as \u201ciOS\u201d and application type as \u201cApp\u201d.<\/p>\n
![\"select](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/select-platform-1.webp\")
<\/a><\/p>\nc.<\/strong> Now, Enter the name of the app and organization identifier, and select SwiftUI interface for building the UI of the app. Also, select Swift as the language for creating the app.<\/p>\n
![\"enter](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/enter-name-of-app-3.webp\")
<\/a><\/p>\nd.<\/strong> Select the folder where you want to save the app and click on Create.<\/p>\n
![\"select](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/select-the-folder.webp\")
<\/a><\/p>\ne.<\/strong> Now your project is ready for development, and you will see something like below.<\/p>\n
![\"project](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/project-is-ready-scaled.webp\")
<\/a><\/p>\nStep 3: Import the ML model into the project<\/h4>\n
Drag and drop the ML model file into the project navigator in Xcode<\/p>\n
![\"import](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/import-the-ml-model-scaled.webp\")
<\/a><\/p>\nStep 4: Create the app UI and functionality<\/h4>\n
In the project navigator, select ContentView.swift
\na.<\/strong> Add the necessary imports and variables: At the top of ContentView.swift<\/strong>, add the following<\/p>\nimport SwiftUI\r\nimport CoreML\r\n<\/pre>\n
b.<\/strong> Inside the ContentView struct, add two @State<\/strong> variables to store the user-entered text and the sentiment prediction:<\/p>\n
@State private var userInput = \"\"\r\n@State private var sentimentPrediction = \"\"\r\n<\/pre>\n
c.<\/strong> Create the UI elements: Inside the body of the ContentView<\/strong> struct, create a VStack<\/strong> that contains a Text view, a TextField view, a Button view, and a Text view to display the sentiment prediction:<\/p>\n
VStack {\r\n Text(\"Sentiment Analysis\")\r\n .font(.largeTitle)\r\n .fontWeight(.bold)\r\n .padding(.top, 50)\r\n Spacer()\r\n VStack {\r\n TextField(\"Enter your text here\", text: $userInput)\r\n .padding()\r\n .background(Color.white)\r\n .cornerRadius(10)\r\n .shadow(radius: 10)\r\n Button(action: {\r\n self.analyzeButton()\r\n }) {\r\n Text(\"Analyze\")\r\n .fontWeight(.bold)\r\n .foregroundColor(.white)\r\n .padding(.vertical, 10)\r\n .padding(.horizontal, 30)\r\n .background(Color.indigo)\r\n .cornerRadius(50)\r\n }\r\n .padding()\r\n Text(sentimentPrediction)\r\n .padding()\r\n .font(Font(UIFont(name: \"Arial\", size: 100)!))\r\n \r\n }\r\n .padding(.horizontal, 30)\r\n .cornerRadius(20)\r\n .padding()\r\n Spacer()\r\n}\r\n<\/pre>\nd.<\/strong> In the Button view, call the analyzeButton()<\/strong> function when the button is tapped:
\nless<\/p>\nButton(action: {\r\n self.analyzeButton()\r\n }) {\r\n Text(\"Analyze\")\r\n .fontWeight(.bold)\r\n .foregroundColor(.white)\r\n .padding(.vertical, 10)\r\n .padding(.horizontal, 30)\r\n .background(Color.indigo)\r\n .cornerRadius(50)\r\n }\r\n .padding()\r\n<\/pre>\ne.<\/strong> Create the analyzeButton()<\/strong> function: Inside the ContentView struct, create the analyzeButton()<\/strong> function that will use the ML model to analyze the user-entered text:<\/p>\n
private func analyzeButton() {\r\n let model = try! SentimentAnalysisClassifier(configuration: MLModelConfiguration())\r\n let input = SentimentAnalysisClassifierInput(text: userInput)\r\n \r\n guard let output = try? model.prediction(input: input) else {\r\n return\r\n }\r\n \r\n sentimentPrediction = output.label\r\n }\r\n<\/pre>\nNow the full code will look like this.<\/strong><\/p>\n
import SwiftUI\r\nimport CoreML\r\nstruct ContentView: View {\r\n @State private var userInput = \"\"\r\n @State private var sentimentPrediction = \"\"\r\n \r\n var body: some View {\r\n VStack {\r\n Text(\"Sentiment Analysis\")\r\n .font(.largeTitle)\r\n .fontWeight(.bold)\r\n .padding(.top, 50)\r\n Spacer()\r\n VStack {\r\n TextField(\"Enter your text here\", text: $userInput)\r\n .padding()\r\n .background(Color.white)\r\n .cornerRadius(10)\r\n .shadow(radius: 10)\r\n Button(action: {\r\n self.analyzeButton()\r\n }) {\r\n Text(\"Analyze\")\r\n .fontWeight(.bold)\r\n .foregroundColor(.white)\r\n .padding(.vertical, 10)\r\n .padding(.horizontal, 30)\r\n .background(Color.indigo)\r\n .cornerRadius(50)\r\n }\r\n .padding()\r\n Text(sentimentPrediction)\r\n .padding()\r\n .font(Font(UIFont(name: \"Arial\", size: 100)!))\r\n \r\n }\r\n .padding(.horizontal, 30)\r\n .cornerRadius(20)\r\n .padding()\r\n Spacer()\r\n }\r\n }\r\n \r\n private func analyzeButton() {\r\n let model = try! SentimentAnalysisClassifier(configuration: MLModelConfiguration())\r\n let input = SentimentAnalysisClassifierInput(text: userInput)\r\n \r\n guard let output = try? model.prediction(input: input) else {\r\n return\r\n }\r\n\r\n\r\n if output.label == \"Positive\" {\r\n sentimentPrediction = \"\ud83d\ude42\"\r\n }\r\n else if output.label == \"Negative\" {\r\n sentimentPrediction = \"\u2639\ufe0f\"\r\n }\r\n else if output.label == \"Neutre\" {\r\n sentimentPrediction = \"\ud83d\ude10\"\r\n }\r\n\r\n\r\n\r\n\r\n }\r\n}\r\nstruct ContentView_Previews: PreviewProvider {\r\n static var previews: some View {\r\n ContentView()\r\n }\r\n}\r\n<\/pre>\n![\"swift](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/swift-sentiment-analysis-app-output-scaled.webp\")
<\/a><\/p>\nStep 5: Test the app on a simulator or a physical iOS device.<\/h4>\n
![\"create](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/sentiment-analysis-swift-output.webp\")
<\/p>\n
![\"sentiment](\"https:\/\/techvidvan.com\/tutorials\/wp-content\/uploads\/sites\/2\/2023\/06\/sentiment-analysis-app-output-2.webp\")
<\/a><\/p>\n