Optimising a [large] look-up table within a loop, in R












0















I would really appreciate any help in optimising the code below. For the number of iterations I need to run, it currently takes way to long.



Basically, I'm looking for the optimal set of model parameters for estimating a variable of interest. I'm simulating a set of model parameters and a variable of interest, using each of them (in a loop) to model the signal of interest, then compare each of the modeled signal to the observed signal, and where they are minimised select one of the (simulated) values we're trying to predict. Then, a RMSE function is used to select which set of model parameters offers the smallest error compared to the observed values.



Many Thanks



Code



#Empty variable

 est_hold <- NULL



# Model Parameters (simulated and/or estimated via NLS earlier)

    beta <- seq(0.001,0.2,0.01)

    sig.g <- seq(0.028, 0.038,length.out = 20)

    sig.v <- seq(0.6, 0.8,length.out = 20)

    params <- data.frame(rbind(beta,sig.g,sig.v))



# Simulated variable of interest (Dependant variable)

sim_var <- seq(1,100,0.1)



# Observed signal (Independant variable)

obs_sig <- seq(0.0001,0.8,length.out = 100000)



# Observed variable of interest (Dependant variable)

obs_var2 <- seq(1,100,length.out = 100000)





min.func <- function(x) {sim_var[which.min(x)]}



system.time({

  for(c in 1:ncol(params)){



    # Modelled signal (Independant) 

    model_var <- params[2,c]*exp(-params[1,c]*sim_var)+params[3,c]*(1-exp(-params[1,c]*sim_var))



    # Calculate difference between each modelled value (using simulated params) and each observed value

    diff <- t(apply(as.data.frame(model_var),1,"-",obs_sig)) 

    diff <- abs(diff)



    # Apply function, to select a simulated dependant value where differences are minimised

    est <- apply(diff,2,min.func)

    rm(diff)



    est_hold <- cbind(est_hold,unlist(est))



  }})



# RMSE function, to calculate RMSE between observed dependant variable and the modelled dependant variable (using simulated params)

rmse.fun = function(x){sqrt(sum((obs_var2-x)^2)/length(obs_var2))}



# Apply the RMSE function to the matrix of modelled values 

# (where each column represents modelled values using each of the simulated params)

sim_rmse <- apply(est_hold,2,rmse.fun)



# Find the position of the lowest RMSE

ind <- which.min(sim_rmse)



#Use this position to subset the parameters (...and continue with script/modelling)

final_params <- params[ind]





r optimization simulation lookup







share|improve this question























  • the data.table package could be interesting for you,..

    – BigDataScientist
    Nov 21 '18 at 22:34











  • I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

    – user2738526
    Nov 22 '18 at 6:00
















0















I would really appreciate any help in optimising the code below. For the number of iterations I need to run, it currently takes way to long.



Basically, I'm looking for the optimal set of model parameters for estimating a variable of interest. I'm simulating a set of model parameters and a variable of interest, using each of them (in a loop) to model the signal of interest, then compare each of the modeled signal to the observed signal, and where they are minimised select one of the (simulated) values we're trying to predict. Then, a RMSE function is used to select which set of model parameters offers the smallest error compared to the observed values.



Many Thanks



Code



#Empty variable

 est_hold <- NULL



# Model Parameters (simulated and/or estimated via NLS earlier)

    beta <- seq(0.001,0.2,0.01)

    sig.g <- seq(0.028, 0.038,length.out = 20)

    sig.v <- seq(0.6, 0.8,length.out = 20)

    params <- data.frame(rbind(beta,sig.g,sig.v))



# Simulated variable of interest (Dependant variable)

sim_var <- seq(1,100,0.1)



# Observed signal (Independant variable)

obs_sig <- seq(0.0001,0.8,length.out = 100000)



# Observed variable of interest (Dependant variable)

obs_var2 <- seq(1,100,length.out = 100000)





min.func <- function(x) {sim_var[which.min(x)]}



system.time({

  for(c in 1:ncol(params)){



    # Modelled signal (Independant) 

    model_var <- params[2,c]*exp(-params[1,c]*sim_var)+params[3,c]*(1-exp(-params[1,c]*sim_var))



    # Calculate difference between each modelled value (using simulated params) and each observed value

    diff <- t(apply(as.data.frame(model_var),1,"-",obs_sig)) 

    diff <- abs(diff)



    # Apply function, to select a simulated dependant value where differences are minimised

    est <- apply(diff,2,min.func)

    rm(diff)



    est_hold <- cbind(est_hold,unlist(est))



  }})



# RMSE function, to calculate RMSE between observed dependant variable and the modelled dependant variable (using simulated params)

rmse.fun = function(x){sqrt(sum((obs_var2-x)^2)/length(obs_var2))}



# Apply the RMSE function to the matrix of modelled values 

# (where each column represents modelled values using each of the simulated params)

sim_rmse <- apply(est_hold,2,rmse.fun)



# Find the position of the lowest RMSE

ind <- which.min(sim_rmse)



#Use this position to subset the parameters (...and continue with script/modelling)

final_params <- params[ind]





r optimization simulation lookup







share|improve this question























  • the data.table package could be interesting for you,..

    – BigDataScientist
    Nov 21 '18 at 22:34











  • I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

    – user2738526
    Nov 22 '18 at 6:00














0












0








0








I would really appreciate any help in optimising the code below. For the number of iterations I need to run, it currently takes way to long.



Basically, I'm looking for the optimal set of model parameters for estimating a variable of interest. I'm simulating a set of model parameters and a variable of interest, using each of them (in a loop) to model the signal of interest, then compare each of the modeled signal to the observed signal, and where they are minimised select one of the (simulated) values we're trying to predict. Then, a RMSE function is used to select which set of model parameters offers the smallest error compared to the observed values.



Many Thanks



Code



#Empty variable

 est_hold <- NULL



# Model Parameters (simulated and/or estimated via NLS earlier)

    beta <- seq(0.001,0.2,0.01)

    sig.g <- seq(0.028, 0.038,length.out = 20)

    sig.v <- seq(0.6, 0.8,length.out = 20)

    params <- data.frame(rbind(beta,sig.g,sig.v))



# Simulated variable of interest (Dependant variable)

sim_var <- seq(1,100,0.1)



# Observed signal (Independant variable)

obs_sig <- seq(0.0001,0.8,length.out = 100000)



# Observed variable of interest (Dependant variable)

obs_var2 <- seq(1,100,length.out = 100000)





min.func <- function(x) {sim_var[which.min(x)]}



system.time({

  for(c in 1:ncol(params)){



    # Modelled signal (Independant) 

    model_var <- params[2,c]*exp(-params[1,c]*sim_var)+params[3,c]*(1-exp(-params[1,c]*sim_var))



    # Calculate difference between each modelled value (using simulated params) and each observed value

    diff <- t(apply(as.data.frame(model_var),1,"-",obs_sig)) 

    diff <- abs(diff)



    # Apply function, to select a simulated dependant value where differences are minimised

    est <- apply(diff,2,min.func)

    rm(diff)



    est_hold <- cbind(est_hold,unlist(est))



  }})



# RMSE function, to calculate RMSE between observed dependant variable and the modelled dependant variable (using simulated params)

rmse.fun = function(x){sqrt(sum((obs_var2-x)^2)/length(obs_var2))}



# Apply the RMSE function to the matrix of modelled values 

# (where each column represents modelled values using each of the simulated params)

sim_rmse <- apply(est_hold,2,rmse.fun)



# Find the position of the lowest RMSE

ind <- which.min(sim_rmse)



#Use this position to subset the parameters (...and continue with script/modelling)

final_params <- params[ind]





r optimization simulation lookup







share|improve this question














I would really appreciate any help in optimising the code below. For the number of iterations I need to run, it currently takes way to long.



Basically, I'm looking for the optimal set of model parameters for estimating a variable of interest. I'm simulating a set of model parameters and a variable of interest, using each of them (in a loop) to model the signal of interest, then compare each of the modeled signal to the observed signal, and where they are minimised select one of the (simulated) values we're trying to predict. Then, a RMSE function is used to select which set of model parameters offers the smallest error compared to the observed values.



Many Thanks



Code



#Empty variable

 est_hold <- NULL



# Model Parameters (simulated and/or estimated via NLS earlier)

    beta <- seq(0.001,0.2,0.01)

    sig.g <- seq(0.028, 0.038,length.out = 20)

    sig.v <- seq(0.6, 0.8,length.out = 20)

    params <- data.frame(rbind(beta,sig.g,sig.v))



# Simulated variable of interest (Dependant variable)

sim_var <- seq(1,100,0.1)



# Observed signal (Independant variable)

obs_sig <- seq(0.0001,0.8,length.out = 100000)



# Observed variable of interest (Dependant variable)

obs_var2 <- seq(1,100,length.out = 100000)





min.func <- function(x) {sim_var[which.min(x)]}



system.time({

  for(c in 1:ncol(params)){



    # Modelled signal (Independant) 

    model_var <- params[2,c]*exp(-params[1,c]*sim_var)+params[3,c]*(1-exp(-params[1,c]*sim_var))



    # Calculate difference between each modelled value (using simulated params) and each observed value

    diff <- t(apply(as.data.frame(model_var),1,"-",obs_sig)) 

    diff <- abs(diff)



    # Apply function, to select a simulated dependant value where differences are minimised

    est <- apply(diff,2,min.func)

    rm(diff)



    est_hold <- cbind(est_hold,unlist(est))



  }})



# RMSE function, to calculate RMSE between observed dependant variable and the modelled dependant variable (using simulated params)

rmse.fun = function(x){sqrt(sum((obs_var2-x)^2)/length(obs_var2))}



# Apply the RMSE function to the matrix of modelled values 

# (where each column represents modelled values using each of the simulated params)

sim_rmse <- apply(est_hold,2,rmse.fun)



# Find the position of the lowest RMSE

ind <- which.min(sim_rmse)



#Use this position to subset the parameters (...and continue with script/modelling)

final_params <- params[ind]





r optimization simulation lookup




r optimization simulation lookup






share|improve this question













share|improve this question











share|improve this question




share|improve this question










asked Nov 21 '18 at 21:00









Sentinel1bSentinel1b

155




155













  • the data.table package could be interesting for you,..

    – BigDataScientist
    Nov 21 '18 at 22:34











  • I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

    – user2738526
    Nov 22 '18 at 6:00



















  • the data.table package could be interesting for you,..

    – BigDataScientist
    Nov 21 '18 at 22:34











  • I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

    – user2738526
    Nov 22 '18 at 6:00

















the data.table package could be interesting for you,..

– BigDataScientist
Nov 21 '18 at 22:34





the data.table package could be interesting for you,..

– BigDataScientist
Nov 21 '18 at 22:34













I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

– user2738526
Nov 22 '18 at 6:00





I'd recommend profvis package as a great way to see what the slowest part of the code is, so you can spend your time trying to fix that slowest bits first

– user2738526
Nov 22 '18 at 6:00












0






active

oldest

votes











Your Answer






StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});

function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});


}
});














draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53420411%2foptimising-a-large-look-up-table-within-a-loop-in-r%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























0






active

oldest

votes








0






active

oldest

votes









active

oldest

votes






active

oldest

votes
















draft saved

draft discarded




















































Thanks for contributing an answer to Stack Overflow!


  • Please be sure to answer the question. Provide details and share your research!

But avoid …



  • Asking for help, clarification, or responding to other answers.

  • Making statements based on opinion; back them up with references or personal experience.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53420411%2foptimising-a-large-look-up-table-within-a-loop-in-r%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown





















































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown

































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown







-

Popular posts from this blog

Biblatex bibliography style without URLs when DOI exists (in Overleaf with Zotero bibliography)

Image
2 2 I have a Zotero bibliography in Overleaf. I use biblatex . usepackage[backend=biber,citestyle=authoryear]{biblatex} addbibresource{zoteroALH.bib} Bibliography entries look like: @book{de_saint-gervais_uniformization_2016, title = {Uniformization of {Riemann} {Surfaces}}, isbn = {978-3-03719-145-3}, url = {https://www.ems-ph.org/books/book.php?proj_nr=198&srch=series%7Chem}, urldate = {2019-02-03}, author = {de Saint-Gervais, Paul Henri}, year = {2016}, doi = {10.4171/145} } In the bibliography down the paper the reference shows like this: There is the DOI, great, but also the URL which is unnecessary when the DOI is there. How can I remove the URL when the DOI is present? biblatex bibliographies overleaf zote
Read more

ComboBox Display Member on multiple fields

Image
2 I'm trying to set the .DisplayMember property of a ComboBox in C#, but I want to bind it to multiple columns in the .DataSouce . My SQL looks like this: SELECT PersNbr, PersFirstName, PersMiddleName, PersLastName FROM Pers WHERE PersNbr = :persNbr; I'm saving this query in a DataTable so each column selected has it's own column in the Datatable . I want to make the .DisplayMember a combination of PersFirstName + PersMiddleName + PersLastName so their full name appears like this: comboBox.DisplayMemeber = "PersFirstName" + "PersMiddleName" + "PersLastName" I know I can just to this on the query: SELECT PersNbr, (PersFirstName || PersMiddleName || PersLastName) PersName and then just do this: comboBox.DisplayMember = "PersName"; but I
Read more

Is it possible to collect Nectar points via Trainline?

Image
2 Since I am not eligible for any Railcard, the only available discount/bonus scheme is Nectar Card for me. GWR, SWR, LNER and Virgin Trains allow to collect Nectar points and I have already linked a card to my GWR account. A friend suggested me to use Trainline App, because it can search for bus tickets either. I had a look at the application, but could not see any option for Nectar. Is there a way to collect Nectar points when I buy tickets via Trainline? uk trains loyalty-programs trainline share | improve this question asked Dec 2 '18 at 15:50 ahmedus 3,224 5 19 49
Read more