javascript time between multiple millisecond dates in an array

Question

If I have an array object of times in milliseconds like this:

  _m7q7sj2wp:
   { '1433891300733': 1433891300733,
     '1433891364133': 1433891364133,
     '1433891526736': 1433891526736,
     '1433891542793': 1433891542793,
     '1433891550861': 1433891550861 },
  _7gh4d0f6l:
   { '1433891352741': 1433891352741,
     '1433891390836': 1433891390836,
     '1433891461772': 1433891461772,
     '1433891523376': 1433891523376,
     '1433891598007': 1433891598007 },

It was created with this:

for (var i = 0, j = results.length; i < j; i++){

    var userguid = results[i].guid;

    var timestamp = results[i].ts;
    if(!guid_t[userguid]){

        guid_t[userguid] = {};

    }

    guid_t[userguid][timestamp] = timestamp;
}

If there is a better way to create the array I'm all for it. I don't really like having to double up the key:value, but I don't know how else to do this.

how would I go about getting the time between each item in the array.

For example, items 1-2, 2-3, 3-4, 4-5.

I'm trying to get the average time between each item.

Fun part is that there are about 400 of these arrays, and each one may have a different amount of items in each.

EDIT: I should mention these are full dates created with +new Date();

Conclusion: I've had time to test all three answers out, and each one worked! (Don't you just love coding?!)

The following tests were using approximately 7000-8000 arrays each. I also had to convert each answer into a function to go through that many, but that addition was trivial and took a few seconds. Then I had to add an additional for loop to combine all the averages into a single value.

Also, please keep in mind this is done with nodejs, with the data coming from mongodb.

These are not the most verbose tests, but they should give people an overview of the performance I had using each answer.

Test 1: Initially, I chose Saumil Soni's answer because it was the first one I tried. The calculations took between 200ms and 1200ms depending on the size of the arrays, and the current server load.

Test 2: Next I tried Jack's answer, and while preforming well, I saw pretty much the same results; sometimes a slight improvment. Calculations were complete in about 175ms to 1100ms.

Test 3: Lastly I tried Keith A's answer. I'm not going to lie, I held off because I didn't understand reduce at all. I did about 10 minutes of reading to learn about it, and it turns out to be dead simple. It's just another way to do a for loop, but uses a different pattern to accomplish it.

This is what I read: Understand Javascript Array Reduce in 1 Minute

So with reduce in my pocket, I tried his answer with a small addition to account for an average not being possible because the array has only one value. Performance was slightly better again, coming in at 75ms to 985ms total

Here is the final code I came to with the help of all three answers.

function avg(sum){

    var avg = sum.reduce(function(total, current, index, array){

        if (index < (array.length-1)){
            return total +  (array[index+1] - current);
        }else{
            return total / (array.length-1);
        }
    }, 0);

    if(isNaN(avg)){

        avg = 30000; //Gave it 30 seconds. Making it 0 lowered the over all average by about 20%. Another option would be to discard it, but I needed it. Change to your liking.
    }

    console.log(avg);

}

In closing I've changed my accepted answer to Keith A's because it worked better in my case. But all three answer are totally amazing!!!

Must the timestamps be unique? Is that why you're adding them like that in the object? — Ja͢ck
– Ja͢ck, Commented Jun 11, 2015 at 7:11
Yes and no. The timestamps are being pulled from mongodb. Each one is in it's own document. — David
– David, Commented Jun 11, 2015 at 7:24
If the uniqueness is already guaranteed, you can use an array. In fact, it's better that way, because object properties are strictly not ordered. — Ja͢ck
– Ja͢ck, Commented Jun 11, 2015 at 7:34

Keith A · Accepted Answer · 2015-06-11 08:00:36Z

2

you could do it using reduce

var sum = [1433891352741,
          1433891390836,
          1433891461772,
          1433891523376,
          1433891598007];

var avg = sum.reduce(function(total, current, index, array){
    if (index < (array.length-1)){
        return total +  (array[index+1] - current);
    }else{
        return total / (array.length-1);
    }
}, 0);


alert("Average: "+ avg + " ms or " + avg / 1000 + " sec");

http://jsfiddle.net/mp1ofx4r/4/

change your last line to

guid_t[userguid].push(timestamp);

so now your guid_t[userguid] contains an array of timestamps instead of objects.

edited Jun 11, 2015 at 8:00

answered Jun 11, 2015 at 7:17

Keith A

8117 silver badges12 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

David Over a year ago

I don't have an array of values like that though; or, how can I change my for loop to create an array like that if there are multiple (read: hundreds) of arrays to create and loop through

Saumil · Accepted Answer · 2015-06-12 15:10:55Z

2

Here is a script which will find the time difference between your arrays and then give you the average interval,

<script>

ar = {
    '1433891300733': 1433891300733,
        '1433891364133': 1433891364133,
        '1433891526736': 1433891526736,
        '1433891542793': 1433891542793,
        '1433891550861': 1433891550861
    };
cal(ar);

function cal(ar) {

    var sum = 0;
    var n = Object.keys(ar).length - 1;
    for (var i = 0; i < n; i++) {
        var x = ar[Object.keys(ar)[i]];
        var y = ar[Object.keys(ar)[i + 1]];
        sum += Math.abs(x-y);
    }

    alert("Average: " + sum / n);
}

I have changed the array name for my readability

EDIT: For accessing the object I have modified my loop so that I can access the object values by their indexes.

edited Jun 12, 2015 at 15:10

answered Jun 11, 2015 at 5:23

Saumil

2,5176 gold badges34 silver badges54 bronze badges

13 Comments

Paul Over a year ago

Why are you calling parseint on numbers? That will just convert them to a string and the back to number.

Saumil Over a year ago

@Paulpro As we have to find the average and perform mathematical operations it is important to convert them into integers because in string format they will just append with each other

Paul Over a year ago

Also its not clear that we wants the absolute difference, instead of the actual difference. I would remove the Math.abs just in case, as it will be easier to notice bugs later where the array elements got out of order.

Paul Over a year ago

There are no strings in the array, only numbers.

Saumil Over a year ago

@Paulpro Well we have to find the average of the difference between the values so I guess abs() is necessary because without it some values would be negative

|

Ja͢ck · Accepted Answer · 2015-06-11 07:24:59Z

Assuming you have a plain array of values:

var values = [1433891300733, 1433891364133, 1433891526736, 1433891542793, 1433891550861];

// calculate differences between V[k - 1] and V[k]
// for 1 <= k < N
var diffs = [],
prev = values[0];

values.slice(1).forEach(function(value) {
    diffs.push(value - prev);
    prev = value;
});

// calculate average difference
var sum = diffs.reduce(function(total, value) {
    return total + value;
});

if (diffs.length) {
    alert(sum / diffs.length);
}

If the numbers must be pulled from an object, you need a few more steps:

var obj = { '1433891300733': 1433891300733,
     '1433891364133': 1433891364133,
     '1433891526736': 1433891526736,
     '1433891542793': 1433891542793,
     '1433891550861': 1433891550861 };

var values = Object.keys(obj).map(function(key) {
  return obj[key];
}).sort();

console.log(values);
// [1433891300733, 1433891364133, 1433891526736, 1433891542793, 1433891550861]

Collectives™ on Stack Overflow

javascript time between multiple millisecond dates in an array

3 Answers 3

1 Comment

13 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

13 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related