lstsq.rst
changeset 132 b8f7ee434b91
child 135 7bc03b5096f9
equal deleted inserted replaced
131:ca42b9821019 132:b8f7ee434b91
       
     1 Hello friends and welcome to the tutorial on Least Square Fit
       
     2 
       
     3 {{{ Show the slide containing title }}}
       
     4 
       
     5 {{{ Show the slide containing the outline slide }}}
       
     6 
       
     7 In this tutorial, we shall look at generating the least square fit line for a
       
     8 given set of points.
       
     9 
       
    10 First let us have a look at the problem.
       
    11 
       
    12 {{{ Show the slide containing problem statement. }}}
       
    13 
       
    14 We have an input file generated from a simple pendulum experiment.
       
    15 
       
    16 It contains two columns of data. The first column is the length of the
       
    17 pendulum and the second is the corresponding time period of the pendulum.
       
    18 
       
    19 As we know, the square of time period of a pendulum is directly proportional to
       
    20 its length, we shall plot l vs t^2 and verify if the proportionality is linear.
       
    21 
       
    22 If it is not linear, we shall generate a least square fit line.
       
    23 
       
    24 {{{ show the slide containing explanation on least square fit }}}
       
    25 
       
    26 As shown in the slide, we are first going to generate the two matrices tsq and
       
    27 A. Then we are going to use the =lstsq= function to find the values of m and c.
       
    28 
       
    29 To read the input file and parse the data, we are going to loadtxt function.
       
    30 Type::
       
    31 
       
    32     data = loadtxt("/home/fossee/pendulum.txt")
       
    33     data
       
    34 
       
    35 As you can see, data is a sequence containing 90 records. Each record contains
       
    36 two values. The first is length and second is time period. But what we need is 
       
    37 two sequences. One sequence containing all the length values and one containing
       
    38 all the time values.
       
    39 
       
    40 Hence we have to use the unpack option with loadtxt. It unpacks the data into
       
    41  sequences depending on the structure of data.
       
    42 
       
    43 Type::
       
    44 
       
    45     l, t = loadtxt("/home/fossee/pendulum.txt", unpack=True)
       
    46     l
       
    47     t
       
    48 
       
    49 We can see that l and t are two sequences containing length and time values
       
    50 correspondingly.
       
    51 
       
    52 Let us first plot l vs t^2. Type::
       
    53 
       
    54     tsq = t * t
       
    55     plot(l, tsq, 'bo')
       
    56 
       
    57 
       
    58 {{{ switch to the plot window }}}
       
    59 
       
    60 We can see that there is a visible linear trend.
       
    61 
       
    62 let us now generate the A matrix with l values.
       
    63 We shall first generate a 2 x 90 matrix with the first row as l values and the
       
    64 second row as ones. Then take the transpose of it. Type::
       
    65 
       
    66     inter_mat = array((l, ones_like(l)))
       
    67     inter_mat
       
    68 
       
    69 We see that we have intermediate matrix. Now we need the transpose.Type::
       
    70 
       
    71     A = inter_mat.T
       
    72     A
       
    73 
       
    74 Now we have both the matrices A and tsq. We only need to use the =lstsq=
       
    75 Type::
       
    76 
       
    77     result = lstsq(A, tsq)
       
    78 
       
    79 The result is a sequence of values. The first item is the matrix p or in simple
       
    80 words, the values of m and c. Hence, ::
       
    81 
       
    82     m, c = result[0]
       
    83     m
       
    84     c
       
    85 
       
    86 Now that we have m and c, we need to generate the fitted values of t^2. Type::
       
    87 
       
    88     tsq_fit = m * l + c
       
    89     plot(l, tsq, 'bo')
       
    90     plot(l, tsq_fit, 'r')
       
    91 
       
    92 We get the least square fit of l vs t^2
       
    93 
       
    94 {{{ Pause here and try out the following exercises }}}
       
    95 
       
    96 %% 2 %% change the label on y-axis to "y" and save the lines of code
       
    97         accordingly
       
    98 
       
    99 {{{ continue from paused state }}}
       
   100 
       
   101 {{{ Show summary slide }}}
       
   102 
       
   103 This brings us to the end of the tutorial.
       
   104 we have learnt
       
   105  * how to use loadtxt to read files
       
   106  * how to generate a least square fit
       
   107 
       
   108 {{{ Show the "sponsored by FOSSEE" slide }}}
       
   109 
       
   110 #[Nishanth]: Will add this line after all of us fix on one.
       
   111 This tutorial was created as a part of FOSSEE project, NME ICT, MHRD India
       
   112 
       
   113 Hope you have enjoyed and found it useful.
       
   114 Thankyou
       
   115  
       
   116 .. Author              : Nishanth
       
   117    Internal Reviewer 1 : 
       
   118    Internal Reviewer 2 : 
       
   119    External Reviewer   :